Expression of Major Capsid Protein VP-1 in the Absence of Viral Particles in Thymomas Induced by Murine Polyomavirus

Abstract

Thymomas induced by polyomavirus strain PTA in mice are known to express the major capsid protein VP-1. Since the expression of a late structural protein such as VP-1 is considered a sign of virus replication, the present work attempted to clarify the implication of the presence of this protein in tumor cells. Electron microscopy of tumors showed a striking absence of viral particles in the vast majority of the cells. However, immunoelectron microscopy of the same samples demonstrated intranuclear VP-1 in most cells despite the absence of viral particles. Very little infectious virus was recovered from tumors. A change in the electrophoretic mobility of VP-1 from thymomas was detected compared with VP-1 from productively infected cells. The data presented in this work prove that the expression of VP-1 in polyomavirus-induced tumors is not synonymous with the presence of infectious virus, suggesting a possible defect in viral encapsidation.

Polyomavirus is a small, nonenveloped, double-stranded DNA virus widely used to study cell transformation in vitro and tumorigenesis in mice (reviewed in reference 4). In vitro, polyomavirus can infect permissive mouse cells, producing infectious virus particles and cell lysis, or transform nonpermissive rat cells. Transformation reflects the complex interaction of viral tumor antigens with key cellular regulators such as the Src family (5, 6, 25, 39), phosphatidylinositol 3-kinase (8, 37, 42), 14-3-3 proteins (7, 33) Shc (10, 23), phosphatase 2A (22, 32), and retinoblastoma protein (19, 24). The genome of polyomavirus encodes early region proteins large T (LT), middle T (mT), and small T (sT) and the late viral structural proteins VP-1, VP-2 and VP-3. During productive infection in mouse cells, both early and late proteins are expressed. LT and sT antigens are important for DNA replication (12, 14, 30, 31), while mT plays a key role in encapsidation through phosphorylation of VP-1 (20, 21). In nonpermissive rat cells only the early antigens are expressed, and mT is the primary viral oncogene (40).

Infection of newborn mice results in a broad tumor distribution. The efficiency of tumor induction depends on both the murine host and the strain of virus used. These are mouse strains that are highly susceptible to tumor induction by polyomavirus, and these include C₃H/BiDa and AKR. Other strains, such as BALB/c or C57BL, are far more resistant. This difference is primarily due to the immune response of mouse strains against the virus (2, 29, 43). Also, some virus strains such as PTA or A2 induce epithelial and mesenchymal tumors involving as many as 14 different cell types within a few months, while others like RA or A3 rarely induce mesenchymal tumors even after as long as a year (9). It has been reported that mT antigens of polyomavirus strains of high or low tumorigenicity are equally effective in their transforming capability, suggesting that other components of the virus account for the difference in tumor formation (16). In this regard, it has been shown that a single amino acid change in the major capsid protein VP-1 is responsible for the difference in the tumor profile, hemagglutination properties, and viral plaque size (17, 18). Various lines of evidence led to the idea that the ability of polyomavirus to induce tumors in mice is directly related to its success in disseminating to different tissues after infection (11, 15). This implies that the cellular receptor for polyomavirus is broadly expressed in mouse tissues. Many attempts were made to characterize this receptor, which is known to bear sialyloligosaccharides that interact differently with high or low transforming polyomavirus strains (1, 3, 35, 36).

Whatever the mechanisms of virus dissemination in mice, it is accepted that polyoma has to first replicate and amplify in several tissues before inducing tumors (17). In C3H Bi/Da mice the highly tumorigenic polyomavirus strain PTA induces mammary, salivary gland, hair follicle, and thymic tumors, and in each tumor, three different cell types coexist. These cell types have been examined for the presence of polyomavirus DNA and the presence or absence of VP-1 (38). The expression of the polyomavirus major structural protein VP-1 in tumor cells implies that virus replication may occur in the tumor (38). However, it has been suggested that, at a single-cell level, viral replication and cell transformation would not be able to coexist (38) because replication would lead to cell lysis. This paradox led us to further characterize virus expression in tumors with a straightforward approach that included the use of transmission electron microscopy (TEM) and immunoelectron microscopy of polyomavirus-induced tumors, together with classic immunocytochemistry and biochemistry. Our results demonstrate the existence of tumor cells where VP-1 is expressed without viral encapsidation. This suggests that the expression of structural viral antigens in tumor cells is not necessarily followed by the synthesis of complete, infectious viral particles.

MATERIALS AND METHODS

Virus.

The polyomavirus PTA strain used was a kind gift of Thomas L. Benjamin (Harvard Medical School, Boston, Mass.). Viral stocks were produced by infecting primary baby mouse kidney cell (BMK) cultures prepared from specific-pathogen-free BALB/c mice obtained from the University of La Plata School of Veterinary Medicine, La Plata, Argentina. Cells were infected at a multiplicity of infection of about 0.1 PFU per cell. Cells were grown in Dulbecco's modified Eagle Medium (DMEM) (Gibco) and 10% fetal calf serum (FCS) (Gibco), then incubated at 37°C in a 5% CO₂ atmosphere until complete cytopathic effect was observed. Cultures were harvested with a rubber policeman and sonicated three times, and cell debris was removed by centrifugation for 15 min at 400 × g. The titers of supernatants were determined on NIH 3T3 cells under agar (41) and then aliquoted and stored at −20°C.

Animals.

Pregnant AKR mice were obtained from the bioterium of the National Academy of Medicine (Argentina) and were kept in individual boxes, fed ad libitum on pellets, and maintained at room temperature with natural periods of light and darkness. Newborn males or females (less than 24 h old) were subcutaneously inoculated with 10⁵ PFU of PTA contained in 0.05 ml of virus stock. Mock-infected mice were inoculated with the supernatant of uninfected BMK cultures. Animals were periodically checked for tumor appearance and sacrificed by ether excess. A complete necropsy of each animal was done, and tumors were dissected and processed immediately as indicated below.

Thymic organotypic cultures.

Thymic tissue from 2-week-old AKR mice was dissected under sterile conditions, minced into 0.5-mm pieces, and placed on sponge strips (Spongostan) in 60-mm-diameter plastic petri dishes. The tissues were fed with DMEM and 15% FCS and infected immediately with 1 ml of polyomavirus stock to achieve a final viral concentration of about 2 × 10⁶ PFU/ml of medium. After 7 days of incubation at 37°C in an atmosphere of 5% CO₂ cultures were harvested, fixed, and embedded for histology and electron microscopy as described below.

Histology and histochemistry.

Tissues were fixed overnight in Bouin fluid. Picric acid was removed by immersion in 70% ethanol, and the samples were dehydrated in ethanol at 96 and 100%, clarified in xylene, and routinely embedded in paraffin. Serial slides were obtained, and hematoxylin-eosin, Masson trichrome, Gomori reticulin, and periodic acid-Schiff staining was performed. Adjacent sections were processed for immunocytochemistry.

Immunocytochemistry.

The peroxidase-antiperoxidase (PAP) method was performed as previously described (26). Briefly, endogenous peroxidase was blocked with 2% hydrogen peroxide in methanol and then washed with 0.05 N Tris-HCl buffer, pH 7.6. To reduce nonspecific background the slides were incubated with 5% normal goat serum in 0.05 N Tris-HCl, pH 7.6. A rabbit polyclonal serum against purified VP-1 obtained by Garcea's method (27) (kindly provided by Thomas L. Benjamin, Harvard Medical School) was used as a primary antibody at a dilution of 1:1,000 in 0.05 N Tris-HCl, pH 7.6. The second and the third sera were, respectively, goat anti-rabbit immunoglobulins (1:50) and rabbit immunoglobulins conjugated with antiperoxidase plus peroxidase (1:250) (Dako). Between incubations, slides were washed thoroughly with 0.05 N Tris-HCl, pH 7.6, and developed under microscopic observation using 0.03% 3-3′diaminobenzidine–2% hydrogen peroxide–0.05 N Tris-HCl, pH 7.6. Slides were slightly counterstained with hematoxylin and then mounted.

Electron microscopy (TEM).

Samples were cut into 0.5-mm-thick pieces and fixed in 4% formaldehyde freshly prepared from paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4, postfixed in osmium tetroxide, and embedded in Vestopal. Slides were obtained using glass knives, and grids were stained with uranyl acetate and lead citrate. Specimens were observed in a Zeiss EM-109-T transmission electron microscope, at 80 kV.

Immunoelectron microscopy.

Samples were fixed as described above and then embedded and frozen in 1.4 M sucrose in PBS and cut in a cryo-ultramicrotome. Slides were prepared in nickel grids and immunocytochemistry was performed using the primary anti-VP-1 serum described above at a dilution of 1:100 in PBS. After washing with PBS, a goat anti-rabbit serum conjugated with colloidal gold (10-nm-diameter particles) was used. Grids were washed again in PBS before viewing. Nonspecific background was blocked by incubating the grids with 5% normal goat serum in PBS at the beginning of the procedure. Slides were counterstained with uranyl acetate.

Virus isolation from animal tissues.

Frozen thymomas were thawed and washed three times with cold PBS and Dounce homogenized in PBS followed by three cycles of freeze-thawing. Debris was removed by centrifugation at 400 × g and discarded. As a control, a polyomavirus-infected BMK monolayer was scraped with a rubber policeman and spun down at 400 × g. Infected cells were collected by centrifugation and processed as described for the thymoma samples. Total protein concentration was measured by the Bradford method. Twenty-five micrograms of protein from each thymoma or BMK infected cells were diluted in 2 ml of DMEM-10% FCS and adsorbed for 2 h at 37°C on NIH 3T3 cells grown to 70% density on several coverslips in 60-mm-diameter plastic petri dishes. As additional controls, NIH 3T3 cells were infected either with 2 ml of virus stock or with 2 ml of mock lysate. Coverslips were fixed at 24 h postinfection (hpi).

For the study of polyomavirus replication in kidneys, homogenates were prepared from tissue sections obtained at necropsy. Homogenates were adsorbed for 2 h to NIH 3T3 monolayers grown on coverslips and then thoroughly washed with DMEM and incubated for 48 h after the addition of DMEM plus 10% FCS. Control cells were adsorbed with a kidney homogenate prepared from uninfected animals. Coverslips were fixed with methanol and processed for immunofluorescence.

Indirect immunofluorescence.

Cells were grown on coverslips and fixed for 30 min in methanol, then washed with PBS, blocked with 5% normal goat serum in PBS, and treated with anti-VP-1 serum at a 1:1,000 dilution in PBS for 1 h at 4°C. After washing thoroughly with PBS, a goat anti-rabbit immunoglobulin conjugated with rhodamine (Sigma) was added at a dilution of 1:100 for 1 h and then washed again with PBS. Coverslips were mounted with 50% glycerol–50% PBS and observed with a Zeiss immunofluorescence microscope with epi-illumination.

Protein electrophoresis and Western blotting.

Protein extracts were obtained from frozen tumors or cell monolayers using a cold extraction buffer (1% NP-40, 0.1% sodium dodecyl sulfate [SDS]–Tris-buffered saline) with protease inhibitors (aprotinin [1 μg/ml], leupeptin [1 μg/ml], pepstatin [1 μg/ml], phenylmethylsulfonyl fluoride [50 μg/ml]) and phosphatase inhibitors (1 mM sodium vanadate and 50 mM sodium fluoride). Other samples were obtained with the same extraction buffer without phosphatase inhibitors. Extracts were then sonicated and proteins were resolved in SDS-acrylamide gels. Gels were blotted onto nitrocellulose membrane and incubated with TNET buffer (10 mM Tris, 3 mM EDTA, 50 mM NaCl, 1% Tween 20 [pH 7.5]) containing 5% nonfat dry milk. For Western blotting, the anti-VP-1 serum was diluted 1:10,000 and used for incubation for 1 h at 4°C, followed by goat anti-rabbit immunoglobulins conjugated with horseradish peroxidase (1:10,000; Sigma). Extensive washing was performed after each incubation using TNET, changing the buffer every 5 min for 1 h. Blots were developed by enhanced chemiluminescence (NEN) using Kodak X-Omat AR film.

RESULTS

Tumor induction, histology, and immunocytochemistry.

The AKR mouse strain was one of the first animal models used in polyomavirus studies, and these mice are known to develop different kinds of tumors after polyomavirus inoculation. In this study, 20 newborn AKR mice were inoculated subcutaneously with polyomavirus, and 15 other newborn mice were mock infected. A third group of 15 newborns was maintained untreated. Twice a week the animals were checked for the presence of tumors. All mice were sacrificed between 9 and 12 weeks postinfection (p.i.), when most of the polyomavirus-infected animals showed breathing difficulty that suggested the presence of intrathoracic masses. While none of the control mice presented any abnormality, 14 of the polyomavirus-infected mice showed tumors at necropsy. Grossly, tumors were yellowish pink, soft, lobulated, noninfiltrating, and large (1.0 to 1.5 cm wide), and all of them were located in the upper mediastinum. Microscopic appearance of tumors showed large neoplastic epithelial cells with acidophilic cytoplasm and clear nuclei with nucleoli and isolated mitotic figures (Fig. 1A). The pattern of reticulin fibers was that of epithelial neoplasms, consisting of single fibers surrounding sheets of cells. Fibrous tissue delimiting lobules and a good vascular support completed the histology of the stroma. No periodic acid-Schiff-positive areas were present. The image corresponded to that of a thymoma. Two animals also showed mammary gland tumors that were not included in this study.

FIG. 1.

Histology and immunostaining of a polyomavirus-induced thymoma. (A) Hematoxylin-eosin staining. Neoplastic epithelial cells with vacuolated nuclei and acidophilic cytoplasms are observed. Necrosis is absent. Magnification, ×100. (B) PAP detection of VP-1 in the same tumor, slightly counterstained with hematoxylin. A high proportion of tumor cells show VP-1-positive nuclei (brown staining). Magnification, ×100.

The PAP method detected the presence of VP-1 in the nuclei of the epithelial cells of the 14 thymomas. The distribution of VP-1 in most tumors was uniform. In these cases an average of one out of five cells showed VP-1 labeling (Fig. 1B). In some other thymomas, patches of VP-1-negative cells were seen alternating with areas of VP-1-positive cells.

Ultrastructural observations and immunoelectron microscopy detection of VP-1.

In order to look for the presence of virus and correlate its presence with VP-1 expression in tumors at the ultrastructural level, thymomas were dissected from mice and immediately fixed at 4°C. Several samples were taken from different parts of each tumor and adjacent tissue for slide preparation and embedding for electron microscopy studies. This approach allowed us to study the same areas with both ultrastructural and histological techniques. Some of the slides obtained from each sample were stained with uranyl acetate and lead citrate to study the characteristics of the tumor and the presence of viral particles. Other slides were used for immunoelectron microscopy using a rabbit anti-VP-1 primary serum, while adjacent slides were treated with normal rabbit serum as a primary. The same procedure was used on polyomavirus-infected thymic organotypic cultures. These cultures were used as a control to elucidate whether polyomavirus can replicate in normal thymus. In order to confirm the specificity of the ultrastructural immunolabeling, polyomavirus-infected and mock-infected BMK cells were harvested at 72 hpi and processed for electron microscopy as described above.

The electron microscopy of thymoma tissues showed uniform cells with branching tonofilaments, abundant polyribosomes, free ribosomes, and elongated cell processes, all characteristic of this kind of neoplasm (34). Viral particles were absent in most tumor cells (Fig. 2 and 3C and 3D). Only a few cells and some intratumoral macrophages showed polyomavirus particles in the nuclei (Fig. 4). In the BMK infected cells, typical polyomaviruses that appeared as 45-nm-diameter, rounded, electron-dense nonenveloped particles (4) were abundant in the nucleus and in the cytoplasm and tightly attached to the cell membranes (Fig. 3A). Virus particles were also observed in the polyomavirus-infected organotypic thymic cultures (Fig. 3B). These observations were done in all 14 thymomas, after embedding two samples of each one and studying at least 100 cells per sample.

FIG. 2.

TEM of a polyomavirus-induced thymoma. No viral particles are observed. (A) There are abundant polyribosomes in the cytoplasm and vacuolated nucleus. The presence of tonofilaments in the cytoplasm indicates the epithelial origin of the tumor. (B) The image shows an elongated cell membrane process typical of a thymoma and two vacuolated nuclei. Magnification, ×20,000.

FIG. 3.

TEM of productively infected monolayers, infected thymic organotypic culture, and polyomavirus-induced thymomas. (A) BMK infected monolayer. Abundant 45-nm-diameter particles are tightly attached to intracytoplasmic cell membranes and free in the cytoplasm, at 96 hpi. Magnification, ×20,000. (Insert) A crystal-like arrangement of polyomavirus. Magnification, ×30,000. (B) Thymic organotypic culture infected with polyoma, 7 days p.i. Viral particles are observed. Magnification, ×20,000. (Insert) A crystal-like arrangement of polyoma in the same tissue. Magnification, ×30,000. (C and D) Nuclei of polyomavirus-induced thymoma cells. No viral particles are observed. Magnification ×10,000.

FIG. 4.

Immunoelectron microscopy of an intratumoral macrophage. Polyomavirus particles are seen exclusively in the nucleus (N) but not in the phagocytic vacuoles (V). Magnification, ×20,000. (Insert) Colloidal gold particles are observed surrounding polyomavirus virions. Magnification, ×55,000.

In about 1 out of 5 to 10 cells from thymomas, VP-1 was ultrastructurally detected by immunolabeling in the nuclei and, to a lesser degree, in the cytoplasm, but in the same fields no polyomavirus particles were observed (Fig. 5D). Heavy labeling was present surrounding polyomavirus particles in BMK infected cultures (Fig. 5B). When normal rabbit serum was used as a primary antibody both in tumors and in the infected BMK cells, labeling was absent (Fig. 5A and C).

FIG. 5.

Immunoelectron microscopy for VP-1 using colloidal gold-conjugated antiserum. Productively infected BMK cells and polyomavirus-induced thymomas are shown. Infected BMK cells were used as a control and treated with normal rabbit serum (A) or with rabbit anti-VP-1 serum (B) as primary antibodies, followed by colloidal gold-conjugated (10-nm-diameter particles) goat anti-rabbit immunoglobulins. No positive labeling is observed in panel A, while colloidal gold particles are observed surrounding polyomavirus virions in panel B. Nucleus of a polyomavirus-induced thymoma cell treated with normal rabbit serum (C) and with anti-VP-1 antibody (D) as primary antibodies followed by colloidal gold-conjugated goat anti-rabbit serum. Strong VP-1 labeling is observed in panel D in the absence of viral particles. Magnification, ×55,000. Bar = 100 nm.

Virus isolation from tumors.

To detect the presence of infectious virus in tumors, the immunofluorescence method described by Türler and Beard (41) was used, since it reflects the infecting titer of polyomavirus stocks and has been used successfully to measure the infectibility of murine cells by polyomavirus (35). Thymoma homogenates were adsorbed onto NIH 3T3 monolayers and after 24 hpi coverslips were fixed in methanol and processed for indirect immunofluorescence. This 24-h infection prevented the spread of the virus from the initially infected cells, allowing a quantitation of the virus contained in the tumors (41). As a control NIH 3T3 cells were incubated with an equal amount of protein extract obtained from BMK infected cells. Only cells that showed clear intranuclear VP-1 staining were scored as productively infected. Tumor extracts from 11 out of 14 thymomas each produced 4 to 20 VP-1-positive cells per 1,000 cells, while extracts of the other 3 tumors did not appear to contain any infectious virus. The infection control presented an average of 596 VP-1-positive cells out of 1,000 cells (data not shown). To confirm that virus was absent in the three negative thymomas, extracts were adjusted to 100 μg/0.1 ml of protein in serum-free DMEM and various dilutions (10⁻¹ to 10⁻⁶) were assayed by plaque assay (41). No plaque forming units were detected even from undiluted samples of the three extracts. A polyomavirus stock produced in BMK cells was quantitated at the same time, yielding a titer of 5 × 10⁶ PFU/ml.

Polyomavirus replication in kidney.

The ability of virus to replicate permissively was explored in mouse kidneys to show that some tissues do allow a viral lytic cycle. Newborn animals were inoculated with polyomavirus PTA as described and two animals were sacrificed at each time point (7, 9, and 12 days p.i.). A second group of mice was mock infected, and two animals from this group were also killed at each time point. Kidneys were obtained at necropsy and studied by microscopy, immunocytochemistry, and TEM. In the example shown in Fig. 6, VP-1 protein was detected in AKR kidneys by the PAP method and adjacent sections examined by TEM showed the presence of viral particles in the renal tubules of polyomavirus-infected mice. This was the case for every time point examined in this experiment. To further confirm the presence of infectious viral particles in the kidneys of the polyomavirus-infected mice, homogenates prepared from the whole organs were adsorbed onto NIH 3T3. At 48 hpi, cells were fixed and processed for immunofluorescence detection of VP-1. Cells that were incubated with kidney homogenates prepared from infected animals showed evidence of extensive lytic viral replication with strong nuclear and cytoplasmic labeling of VP-1. No VP-1 was detected either by PAP staining of uninfected kidney tissues or by immunofluorescence of NIH 3T3 cells exposed to homogenates of the uninfected tissues. These data agree with those of others (11) that described polyomavirus replication in kidneys of C₃H/BiDa mice.

FIG. 6.

Kidney of animal infected with polyomavirus. (A and B) PAP staining for polyomavirus VP-1 without counterstaining with hematoxylin. Positive nuclear immunolabeling of cells is observed in the tubules. Magnification, ×100 (A) and ×200 (B). (C) High magnification of a kidney cell from a collector tubule shows intranuclear polyomavirus particles. Magnification, ×45,000.

VP-1 electrophoresis.

Next, characteristics of the VP-1 protein expressed in thymomas were examined. Frozen sections of tumors and productively infected NIH 3T3 cells were extracted and used for immunoblot detection of VP-1 using anti-VP-1 polyclonal serum. All 14 tumors showed a 45-kDa protein that comigrated with VP-1 from control infections when resolved in an SDS–12% polyacrylamide gel electrophoresis (SDS–12% PAGE) gel (Fig. 7A), even the three tumors that showed no infectious virus in the titration assays. The amount of total protein loaded for each sample was adjusted to achieve a similar VP-1 signal in each lane of the blot, because of the considerable variation in the expression of VP-1 in thymomas from different animals. Since it has been reported that VP-1 undergoes a posttranslational phosphorylation that is required for efficient encapsidation of the viral particles (such as threonine phosphorylation), the phosphorylation state of VP-1 found in the tumors was examined and compared to that in productively infected cells. In these studies VP-1 protein was extracted using phosphatase inhibitors in the extraction buffer. Proteins were resolved using an SDS–17% PAGE gel followed by immunoblotting of VP-1 as described above. Western blots revealed that, under these conditions, the mobility of the VP-1 present in tumors was faster than that of VP-1 from NIH 3T3 cells (Fig. 7B). When the same experiment was done with samples extracted without phosphatase inhibitors, VP-1 from all the samples showed the same mobility (Fig. 7C).

FIG. 7.

VP-1 protein in thymomas and productively infected cells. Western blot analysis was used to detect VP-1 protein in thymoma homogenates. (A) VP-1 Western blot after standard protein extraction and electrophoresis on an SDS–12% PAGE gel. Lanes: 1, polyomavirus-infected NIH 3T3 extract, 2, normal thymus from an uninfected mouse; 3 to 5, thymomas from three different animals. (B) VP-1 Western blot after extraction using phosphatase inhibitors and electrophoresis on an SDS–17% PAGE gel. VP-1 proteins present in thymomas (lanes 1, 3, and 5) show a faster electrophoretic mobility than VP-1 from productively infected NIH 3T3 cells (lane 4). Lane 2 contains uninfected NIH 3T3 extract. VP-1 from productively infected NIH 3T3 cells was loaded in the gel between two samples from thymomas in order to better show the shift. (C) Western blot after extraction without using phosphatase inhibitors and electrophoresis on an SDS–17% PAGE gel. VP-1 from productively infected NIH 3T3 cells (lane 2) shows the same electrophoretic mobility as VP-1 proteins present in thymomas (lanes 1 and 3). The position of a molecular mass marker (in kilodaltons) is indicated to the right of each panel.

DISCUSSION

This work has focused on the presence of the major virus capsid protein VP-1 in virus-induced tumors and the possible implications for virus replication. Anatomical, histological, histochemical, and ultrastructural findings with the tumors that arose in mice inoculated with polyomavirus PTA were consistent with thymomas, which are neoplasms developed from the epithelial component of the thymus. Previously, three different kinds of cells were described to coexist in polyomavirus-induced tumors. They were classified according to the VP-1 and viral DNA content. Type 1 contained unintegrated viral DNA and the structural viral protein VP-1; type 2 only contained viral DNA without simultaneous expression of VP-1; and type 3 had only a few copies of polyoma DNA possibly integrated into the cell chromosomes and no VP-1 (38). The presence of VP-1 in type 1 cells has been considered to be the result of virus replication in tumors (38). Based on these data, our first approach was to study by TEM the distribution of virus particles in these polyomavirus-induced tumors and the spread of polyomavirus infection from cell to cell. The most striking observation was that the vast majority of the tumor cells did not contain viral particles. The experiment included a TEM study of over 100 cells from every suitable tumor sample. Very few isolated cells showed intranuclear viruses. PAP labeling of these same tumors demonstrated that an average of one in five tumor cells was VP-1 positive in each tumor. However, some areas of the tumor tissues were VP-1 negative. To ensure that VP-1-positive cells were evaluated for the presence of virions, immunoelectron microscopy was performed using the same anti-VP-1 antibody as the one used in the PAP labeling. As before, about 1 of every 5 to 10 cells contained VP-1 in their nuclei or, to a lesser extent, cytoplasms but no polyomavirus particles were observed in these cells. In contrast, infected BMK cells showed strong and specific labeling surrounding each viral particle. This result suggested that in most VP-1-positive thymoma cells, the late region of the genome was expressed but, somehow, viral encapsidation failed. It might be argued that normal thymic tissue would not allow viral encapsidation and production of infectious viral particles. To resolve this point, thymic organotypic cultures from 2-week-old AKR mice were infected in vitro with the PTA strain of polyomavirus. After 7 days of incubation, VP-1 could be demonstrated by the PAP method in paraffin-embedded slides, and typical polyomavirus particles were observed by TEM in the epithelial component of the normal thymus. Thus, the absence of virus particles in thymomas is not due to an intrinsic inability of the thymic epithelium to support polyomavirus replication but, instead, to other factors present solely in the neoplastic thymic tissue. Furthermore, here we show that lytic infection of the virus occurs in mouse kidneys at early times p.i. but this organ never develops neoplasms as a result of this process. Moreover, unpublished data form our laboratory show that lytic infection is also present at this early stage of polyoma infection in organs such as mammary gland and parotid glands which do develop tumors some 2 to 3 months later. Thus, the relationship between polyomavirus replication and tumor induction is still far from clear and seems to be tissue specific. Possibly in some organs, at a certain stage during the virus infection there is a switch between lytic replication and nonproductive infection which results in tumor development.

The very low amount of infectious virus in the thymomas was confirmed by adsorption on permissive cell monolayers and VP-1 detection by immunofluorescence. Eleven tumors showed only 4 to 20 VP-1-positive cells per 1,000 compared with the productively infected control that showed 596 positive cells out of 1,000. Moreover, in three thymomas no VP-1-positive cells were seen at all, and this result was confirmed by direct titer determination. If the large number of VP-1 positive cells detected immunocytochemically (Fig. 1B) had been virus producing, far more infectious virus should have been detected in these assays. Overall, these results support the idea that most of the thymoma cells do not contain infectious virus and is in agreement with other reports indicating little recovery of infectious viruses from tumors (44).

Where did the small amounts of virus detected in the thymomas come from? There are two plausible explanations: first, we have previously demonstrated that polyomavirus can replicate in macrophages obtained from mouse peritoneum, spleen, and lung (unpublished data). Since macrophages are present in normal and neoplastic thymus, the virus obtained from the tumors may arise from these cells. In this study, we did detect macrophage-like cells in thymomas, and these had polyomavirus particles only in the nuclei, not in the vacuoles, thus suggesting that macrophages were productively infected. Second, a low number of thymoma cells might allow polyomavirus replication at a given point in time, a semipermissive situation. In this study, viral particles were detected by TEM in a few isolated cells, but this was not a frequent phenomenon. Polyomavirus replication in the intratumoral macrophages or in isolated thymoma cells could then explain the low presence of infectious virus in thymoma homogenates.

The data described in this report demonstrate that the expression of VP-1 in tumor cells is not synonymous with the presence of infectious viral particles. Various molecular explanations might explain this discrepancy. One is that polyomavirus genomes present in thymoma cells may have partial deletions involving VP-1 coding sequences. It is known that for VP-1 to be synthesized all the early genes (LT, mT, and sT) must be expressed, but small, internal deletions involving VP-1 or VP-2/3 coding sequences could prevent viral encapsidation. In fact, some of these deletions are known to exist in several polyomavirus-induced tumors (although not in thymomas) (38) and have also been described in polyomavirus-induced mammary gland tumors by viral DNA and RNA studies (45). If deletions are present, VP-1 immunolabeling would probably not be altered since the primary serum used in this work is a polyclonal antibody against purified VP-1. Another possibility is that the intact viral genome is expressed in most thymoma cells but encapsidation does not take place. VP-1 has different isospecies, which require posttranslational modifications (13) such as threonine phosphorylation in order to assemble (21, 28). Thus, it is possible that some posttranslational modifications are not present in the VP-1 synthesized in the thymoma cells. In support of this possibility, VP-1 extracted from thymomas showed slightly different migration on SDS-PAGE gels and Western blots than the protein prepared from productively infected cells. It is interesting that this shift was observed only when the VP-1 extraction was done with buffers that avoided phosphatase activity but was not evident when standard extraction methods were used. Although these data do not completely rule out small or single-amino-acid deletions in VP-1, they are more consistent with the possibility that a difference in posttranslational modification of VP-1 may be involved in the defective viral assembly observed in tumors. It is not possible in whole animal experiments to study VP-1 phosphorylation with the classic ³²P pulse-chase and thin-layer chromatography method. Thus, a more elaborate approach will be necessary to fully define the phosphorylation state of VP-1 in these tumors, which is beyond of the scope of this paper.

The results described in this work support the fact that VP-1 expression is not necessarily followed by virus encapsidation in polyomavirus-induced thymomas. Thus, these tumors include a fourth cell type in addition to the three already described in polyomavirus-induced tumors: a cell where the major virus capsid protein is synthesized but where infectious virus particles are absent. These data would require revised consideration of polyomavirus replication in tumors and the nature of virus spread to other organs. In this regard, a recently published paper used a completely different approach and proposed that dissemination of polyomavirus in mice occurs as free virus DNA and not as encapsidated viruses (45). This independent observation reinforces the data presented here. Hopefully these results will allow a better understanding of the complex relationship between polyomavirus replication and tumor induction in mice.