Emergent Human Pathogen Simian Virus 40 and Its Role in Cancer

The discovery of the polyomavirus SV40, as well as its introduction as a pathogen into the human population, was tied to the development and worldwide distribution of early forms of the polio vaccine (13, 95, 111, 123). Inactivated (Salk) and early live attenuated (Sabin) forms of polio vaccines were inadvertently contaminated with SV40 (95, 97, 111). In addition, different adenovirus vaccines distributed to some U.S. military personnel from 1961 to 1965 also contained SV40 (64). The viral contamination occurred because these early vaccines were prepared in primary cultures of kidney cells derived from rhesus monkeys, which are often naturally infected with SV40 (13, 95, 111). Infectious SV40 survived the vaccine inactivation treatments, and conservative estimates indicate that up to 30 million people (children and adults) in the United States may have been exposed to live SV40 from 1955 through 1963 when administered potentially contaminated polio vaccines (95, 111). Millions of people worldwide were also potentially exposed to SV40 because contaminated polio vaccines were distributed and used in many countries (85, 123). These data led the Institute of Medicine to conclude that “the biological evidence is of moderate strength that SV40 exposure from the polio vaccine is related to SV40 infection in humans” (111).

Shortly after its discovery, SV40 was shown to be a potent oncogenic DNA virus (13). In animal models, the neoplasias induced by SV40 included primary brain cancers, malignant mesotheliomas, bone tumors, and systemic lymphomas (13). Subsequently, many in vitro studies established that the oncogenic capacity of SV40 reflects the disruption of critical cell cycle control pathways (9, 96, 116). During the last decade, numerous published studies from independent laboratories, using different molecular biology techniques, have demonstrated SV40 large tumor antigen (T-ag) or DNA in primary human brain and bone cancers and malignant mesothelioma (1, 13, 39, 50, 123). More recently, studies have demonstrated that SV40 T-ag sequences are significantly associated with non-Hodgkin's lymphoma (NHL) (102, 124, 125). Therefore, the major types of tumors induced by SV40 in laboratory animals are the same as those human malignancies found to contain SV40 markers. A recent meta-analysis (122) of the molecular evidence conclusively established a significant excess risk of SV40 with those selected human cancers.

It is noteworthy that SV40 has been detected in malignancies from children and young adults not exposed to contaminated polio vaccines, as well as in older adults (5, 18, 71, 73, 76, 117, 124, 125, 129, 132, 133). The detection of viral markers in young persons, by using molecular techniques, coupled with the isolation of infectious SV40 from tumors (62) and from nonneoplastic specimens (66, 67), suggests that SV40 continues to cause infections in the human population today. In contrast, some retrospective epidemiological studies have failed to demonstrate an increased cancer risk in populations which had a high likelihood of having received potentially contaminated polio vaccine (20, 82, 95, 112, 114). However, the epidemiological data available are recognized to be inconclusive and limited (95, 111, 123), and the Institute of Medicine found that the epidemiological data for cancer rates in people potentially exposed to SV40-contaminated vaccines are inadequate to evaluate a causal relationship (111). This conclusion is based on the lack of data on which individuals actually received contaminated vaccines, the unknown dosage of infectious SV40 present in particular lots of vaccine, the failure to know who among the exposed were successfully infected with SV40, the inability to know if the vaccine “unexposed” cohorts may have been exposed to SV40 from other sources, and the difficulty of monitoring a large population for cancer development for years after virus exposure. These important limiting factors led the Institute of Medicine to “not recommend additional epidemiological studies of people potentially exposed to contaminated polio vaccine.”

SV40 is in the family Polyomaviridae, which includes JC virus (JCV) and BK virus (BKV). Polyomaviruses are small, nonenveloped, icosahedral DNA viruses. Their genomes consist of a single copy of double-stranded, circular, supercoiled DNA about 5 kb in length. BKV and JCV share 72% DNA sequence homology, and each shares approximately 70% homology with SV40. Although these viruses are related, they can be distinguished easily at the DNA and protein levels. Genetic differences, particularly in the noncoding, regulatory regions of the viral genomes, may determine important differences in host range. Furthermore, the three viruses can be differentiated serologically by neutralization and hemagglutination assays (52, 56, 98).

The SV40 genome is divided into early and late regions, with the early region coding for the large and small T-ags and the late region encoding the capsid proteins VP1, VP2, and VP3 (Fig. 1). Large T-ag of SV40 strain 776 contains 708 amino acids and is a very multifunctional protein (Fig. 2). The large T-ag is an essential replication protein that is required for initiation of viral DNA synthesis and that also stimulates host cells to enter S phase and undergo DNA synthesis. Because of this ability to subvert cell cycle control, T-ag represents the major transforming protein of SV40. T-ag forms complexes with several cellular proteins; fundamental to T-ag effects on host cells is binding to cellular tumor suppressor proteins (9, 13, 96, 116). These properties help explain SV40's potential as a tumor virus. However, it is important to point out that the oncogenic capacity of SV40 is an accidental side effect of the viral replication strategy; viral proteins (large and small T-ags) in lytically infected cells stimulate host cells into a state capable of supporting viral replication.

There is only one known serotype of SV40, but genetic strains exist and can be distinguished by nucleotide differences in the regulatory region (60) and in the variable domain at the extreme C terminus of T-ag, which is defined as the last 86 amino acids of the molecule (residues 622 to 708) (47, 58, 62, 63, 81, 109, 110). Nucleotide differences in the T-ag C-terminal region, including polynucleotide insertions and deletions as well as single nucleotide changes, would change some encoded amino acids. These distinctions at the nucleotide and protein levels have conclusively established that SV40 sequences in human malignancies and other clinical samples are not the result of accidental laboratory contamination (Fig. 3, 4, and 5). However, future studies need to determine whether SV40 strains differ in pathogenic and/or oncogenic capacity. The classic example of DNA virus strains differing in oncogenic capacity is the human papillomavirus group; of the more than 75 types described, of which about 30 cause genital infections, only a few types are associated with the development of cervical carcinoma (68, 135). This identification of high-risk strains has led to the development of preventive interventions, such as the vaccine against human papillomavirus type 16 (54).

Laboratory-adapted monkey strains of SV40 typically contain two 72-bp enhancer elements (Fig. 4 and 5); these are designated “nonarchetypal” or complex regulatory structures (60). In contrast, SV40 isolates from human nonmalignant (Fig. 4) and malignant (Fig. 5) specimens usually (but not always) contain no duplications in the enhancer (“archetypal” structure).

Although the function of the SV40 T-ag variable domain is not known, experimental data have suggested that it may be important in some aspect of the virus-host interaction. Embedded within the variable domain of large T-ag is a functional domain, which encompasses amino acids 682 to 708, defined as the host range/adenovirus helper function (hr/hf) domain (Fig. 2). A C-terminal fragment of T-ag can relieve the adenovirus replication block in monkey cells (23, 41, 51, 90, 106) by an unknown mechanism. The hr function was identified because T-ag C-terminal deletion mutants exhibited different growth properties in monkey cell lines; the deletion mutants grew very poorly in CV-1 cells but grew well in BSC and Vero cells (24, 87, 118, 119). Viral DNA was replicated to near-wild-type levels in all three cell types (87, 108). Virions produced by the hr/hf mutants do not assemble properly, seemingly due to an inability to add VP1 to the 75S assembly intermediates (105).

The functional roles for another SV40 early protein, small T-ag, are more elusive. This protein is not essential for virus replication in tissue culture, and there is not a uniform requirement for it in SV40 transformation or tumor induction. However, studies indicate that SV40 small T-ag enhances large T-ag-mediated transformation (96) and is required for complete transformation of human cells in vitro (42). It inhibits cellular protein phosphatase 2A by complexing with the catalytic subunit and regulatory subunit of the enzyme. Small T-ag plays a role in the induction of telomerase in SV40-infected human mesothelial cells (36). Also, recent data indicate that small T-ag is required by large T-ag to upregulate Notch1 expression in SV40-infected and -transformed human mesothelial cells, as well as in SV40-positive human mesotheliomas (7).

An appreciation of the replication cycle of SV40 is fundamental to understanding the oncogenic capacity of SV40 and its potential etiologic role in some human malignancies. The major histocompatibility class I molecules are the specific cell surface receptors for SV40 (4, 8). This initial step in the viral cycle helps explain the broad tropism of the virus and its ability to infect and induce transformation in many types of cells and tissues. In addition, it provides an important distinction between SV40 and the other two polyomaviruses that are able to infect humans, JCV and BKV. JCV uses an N-linked glycoprotein and BKV uses a glycolipid as unique host cell receptors (3). These marked differences are believed to affect the nature of infections by these three viruses in tissues and individuals.

After infection of a cell, SV40 produces large and small T-ags early in the viral replication cycle. These antigens bind and block important tumor suppressor proteins, which include p53, pRb, p107, and p130/Rb2 (1, 13, 59, 96) (Fig. 2). The functions of these intracellular proteins are centered in the control of the cell cycle. The tumor suppressor p53 is believed to sense DNA damage and either pauses the cell in late G₁ for DNA repair or directs the cell to commit suicide through the apoptotic pathway (96, 116). SV40 T-ag binding sequesters p53, abolishing its function and allowing cells with genetic damage to survive and enter S phase, leading to an accumulation of T-ag-expressing cells with genomic mutations that may promote tumorigenic growth. pRb normally binds transcription factor E2F in early G₁ of the cell cycle; T-ag causes unscheduled dissociation of pRb/E2F complexes, releasing E2F to activate expression of growth-stimulatory genes (96, 116). Therefore, SV40 infections in humans may interfere with several pathways related to cell cycle control and lead to development of malignancies.

Studies indicate that SV40 can replicate productively in human cells, including fetal tissues (101), newborn kidney cells (101), and different human tumor cell lines (83), although it grows poorly in human fibroblasts (84). Moreover, in vitro assays have shown that human cells can support replication of SV40, establishing that human proteins have the intrinsic ability to cooperate with SV40 T-ag to replicate SV40 DNA (65, 80, 127). Some human cell types undergo visible cell lysis in response to SV40, whereas others fail to exhibit cytopathic changes and produce low levels of virus (84). General conclusions from these early studies are that SV40 can replicate in human cells and that various human cell types display differences in susceptibility to infection by SV40. The basis for the differences is unknown, but T-ag functions are believed to be important (27, 69).

Recent studies have shown that primary human mesothelial cells respond to SV40 very differently from fibroblasts; the mesothelial cells are highly susceptible to SV40 infection and transformation. Most mesothelial cells were infected; few were killed; high levels of p53/T-ag complexes were present; Notch1, the hepatocyte growth factor receptor (Met), and insulin-like growth factor 1 were upregulated; and the tumor suppressor gene RASSF1A was inhibited (6, 15, 39, 93). SV40-positive human mesotheliomas show similar changes. The rate of transformation of SV40-infected mesothelial cells was at least 1,000 times higher than that of human fibroblasts (6). These findings emphasize that different human cell types may display dramatically different virus-cell interactions during infection.

The recognized natural hosts for SV40 are species of Asian macaque monkeys, especially the rhesus macaque (Macacca mulatta). SV40, like the polyomaviruses JCV and BKV, establishes persistent infections, often in the kidneys of susceptible hosts (13, 59). An association of primary polyomavirus infection with mild respiratory tract disease, mild pyrexia, and transient cystitis has been reported (32), but the route of infection of these three viruses has not been firmly defined.

SV40 infections may become latent, and the level of virus present may be very low. Both viremia and viruria occur in infected animals, and virus shed in the urine is the probable means of transmission (2, 97). SV40 infections in healthy monkeys appear to be asymptomatic (100), but SV40 causes widespread infections among monkeys that are immunocompromised due to simian immunodeficiency virus infection (47, 58, 81); SV40 sequences and infectious virus were detected in brain, kidney, spleen, and peripheral blood mononuclear cells (PBMCs). These results demonstrate that SV40 can be an opportunistic pathogen in immunosuppressed hosts and that the virus may spread within the host by hematogenous routes.

The oncogenic capacity of SV40 infections has been well established in laboratory animal models (9, 13, 19, 111, 123). The latent period of tumor development in hamsters infected with SV40 ranges from 3 months to more than a year. The frequency of tumor development is usually over 90% in animals infected as newborns but is reduced in older animals. These data suggest that the age at the time of infection, the route of infection, and the duration of infection may be factors influencing the development of malignancies by SV40.

The neoplasias induced by SV40 in animal models include primary brain cancers, malignant mesotheliomas, bone tumors, and systemic lymphomas (13, 39, 123). Lymphomas are a common malignancy during SV40 infection. In hamsters inoculated intravenously with SV40, systemic lymphomas developed among 72% of the animals, compared to none in the control group (21, 29, 30). The lymphomas were of B-cell origin (22). Following intravenous inoculation, about one-third of the animals developed more than one histologic type of neoplasm, with osteosarcomas being most common after lymphomas. Following intracardiac inoculation, malignant mesotheliomas and osteosarcomas developed in addition to lymphomas (19). An etiologic role of the virus in those cancers was supported because SV40 T-ag was expressed in all malignant cells, animals with tumors developed antibody against SV40 T-ag, and neutralization of SV40 with specific antibody before virus inoculation prevented cancer development (29, 30). Knowledge of these models prompted us, as well as other investigators, to consider the role of polyomavirus SV40 infections in some human malignancies.

During the last decade, many studies have shown the presence of SV40 large T-ag DNA or other viral markers in primary human brain and bone cancers, malignant mesotheliomas, and NHL (Fig. 6). Sequence analyses (Fig. 3 and 5) and detection of T-ag protein (Fig. 7) ruled out laboratory contamination of tumor samples. Importantly, infectious SV40 was isolated from a primary brain cancer of a 4-year-old child (62). An important consideration when evaluating the molecular biology data is the sensitivity of methods used to detect SV40 in human tumor samples. Early studies (before 1992) identified SV40-positive neoplasms by using indirect immunofluorescence for viral proteins or DNA hybridization techniques (55, 74, 130), whereas studies after 1992 generally used PCR-based assays.

During the last three decades more than 60 original studies have reported the detection of SV40 in primary brain and bone cancers, malignant mesothelioma, and NHL, whereas a few studies have described an absence of SV40 in those malignancies (16, 33, 34, 44, 48, 70, 113). However, the small numbers of samples tested, the histologic types of malignancies examined, and the laboratory methodologies employed in some cases limit the significance of the results in those studies reported to be negative. Indeed, several steps need to be considered when performing molecular studies of human specimens (1, 50, 61, 107). First, the extraction step of nucleic acids determines whether tissues yield adequate and suitable DNA or RNA for analysis. Unfortunately, with formalin-fixed and paraffin-embedded specimens, degradation of nucleic acids and proteins is a common problem, and the quality of recovered DNA may be poor. If only small amounts of paraffin-embedded tissues are available, the yield of nucleic acids may be inadequate for analysis. Primers directed to a human cellular gene should be used to establish the suitability of a sample for PCR analysis. Because of the sensitivity of PCR-based assays, it is important to rigorously guard against laboratory contamination of samples and controls during processing or testing. Tissue processing and PCR assay setup should be performed in different facilities, from which positive controls (i.e., plasmids) are excluded. Negative tissue controls, extracted and analyzed in parallel, should be included in each experiment to monitor for reagent contamination. The selection of primers and PCR conditions greatly influences the sensitivity and reliability of the assay. Another factor is that tumor specimens usually contain mixtures of normal and malignant cells, in varying proportions. Variations in one or more of these important parameters may explain, at least in part, the ranges in positivity observed among some positive studies and the results obtained in some negative studies.

Table 2 provides a timeline for landmark discoveries associating the polyomavirus SV40 and human malignancies. Although numerous studies have detected SV40 in human primary brain and bone cancers, malignant mesothelioma, and NHL, the small sample sizes and the lack of a control group in some studies made it difficult to make conclusions about the extent to which SV40 may be associated with those human cancers. For this reason, we conducted a meta-analysis of controlled studies (122), an approach which can provide a more balanced and less biased estimate of the evidence than individual studies (57). For inclusion in the meta-analysis, reports had to meet the following criteria: studies were conducted among patients with primary malignancies, the investigation of SV40 was performed on primary cancer specimens and not on cultured cells, the analysis included a control group, and the same laboratory technique was used for both case and control samples. These criteria were established because the use of appropriate controls is crucial in the proper analysis of tissue for viral DNA, especially considering the sensitivity of PCR techniques (38). Thirty-five independent studies met these inclusion criteria. In total, data from 1,793 patients with primary malignancies were evaluated to determine whether SV40 is significantly associated with primary brain cancer, malignant mesothelioma, bone cancer, and NHL.

Thirteen studies fulfilled the criteria for the investigation of primary brain cancers (Table 3). The combined odds ratio (OR) of the studies used in the analysis was 3.9 (95% confidence interval [CI], 2.6 to 5.8). This effect was based on specimens from a total of 1,143 patients, of which 661 were primary brain cancer samples and 482 were control specimens. A modifier detected was the type of sample analyzed (paraffin embedded versus frozen). The adjusted OR was 3.8 (95% CI, 2.6 to 5.7). For malignant mesothelioma, 15 studies fulfilled the criteria; the combined OR of analysis was 16.8 (95% CI, 10.3 to 27.5) and was based on 528 patients with malignant mesothelioma and 468 controls (Table 4). Modifiers detected were the type of control tissue and the method of detection of SV40. The adjusted OR was 15.1 (95% CI, 9.2 to 25.0). The combined OR of the analysis of bone cancers and SV40 was 24.5 (95% CI, 6.8 to 87.9) and was based on 303 patients with bone tumors and 121 controls from four reports (122). The OR for NHL was 5.4 (95% CI, 3.1 to 9.3) and represented 301 cases and 578 controls included in three studies (Table 5). Because there were only three studies that fulfilled the inclusion criteria, further examination of modifying variables was not possible for NHL.

This analysis of published reports found a significant excess risk of SV40 associated with human primary brain cancers, malignant mesotheliomas, bone cancers, and NHL compared to control samples. Therefore, the major types of human malignancies associated with SV40 are the same as those induced by SV40 in animal models. Although the proportion of human cancers containing SV40 varied from study to study, viral prevalence was always greater among primary tumors than among control tissues. Importantly, analysis of data indicated that SV40 may be etiologically meaningful in the development of a specific subset of human cancers. Multiple studies have shown the expression of SV40 mRNA and/or T-ag in cancer cells, the integration of SV40 sequences in some cancers, and SV40 T-ag protein complexed with p53 and pRb in some tumor specimens (1, 10, 13, 39, 50, 76, 122). These findings are compatible with current understanding of how SV40 T-ag mediates oncogenesis. Moreover, microdissection of human malignant mesothelioma samples followed by PCR detected SV40 T-ag DNA only in cancer cells and not in adjacent nonmalignant cells (1, 39, 104). These results from different experimental studies support the conclusion of the Institute of Medicine (111) that “the biological evidence is of moderate strength that SV40 exposure could lead to cancer in humans under natural conditions.”