OUP user menu

Potential Mechanisms of the Human Polyomavirus JC in Neural Oncogenesis

Luis Del Valle MD, Martyn K. White DPhil, Kamel Khalili PhD
DOI: http://dx.doi.org/10.1097/NEN.0b013e318180e631 729-740 First published online: 1 August 2008


The human polyomavirus JC (JCV) is a small DNA tumor virus and the etiologic agent of the progressive multifocal leukoencephalopathy. In progressive multifocal leukoencephalopathy, active JCV replication causes the lytic destruction of oligodendrocytes. The normal immune system prevents JCV replication and suppresses the virus into a state of latency so that expression of viral proteins cannot be detected. In a cellular context that is nonpermissive for viral replication, JCV can affect oncogenic transformation. For example, JCV is highly tumorigenic when inoculated into experimental animals, including rodents and monkeys. In these animal tumors, there is expression of early T-antigen but not of late capsid proteins, nor is there viral replication. Moreover, mice transgenic for JCV T-antigen alone develop tumors of neural tube origin. Detection of JCV genomic sequences and expression of viral T-antigen and agnoprotein suggest a possible association of this virus with a variety of human brain and non-CNS tumors. Here, we discuss the mechanisms involved in JCV oncogenesis, briefly review studies that do and do not support a causative role for this virus in human CNS tumors, and identify key issues for future research.

Key Words
  • β-Catenin
  • Brain tumor
  • Human polyomavirus JC insulin-like growth factor pathway
  • Progressive multifocal leukoencephalopathy
  • T-Antigen


Polyomaviruses are a family of nonenveloped DNA tumor viruses with icosahedral capsids that infect many vertebrate species, including birds, rodents, and primates (1). The polyomavirus genome is composed of circular, closed, and supercoiled DNA; it is small in size (~5,000 base pairs) and is composed of early and late regions that are controlled by a common noncoding regulatory region (NCCR) that lies between them. This region represents a bidirectional promoter from which early and late transcription are initiated in opposite directions (1). Until recently, 3 species of polyomavirus were known to infect humans: SV40, BK virus (BKV), and human polyomavirus JC (JCV). SV40, the natural host of which is the rhesus macaque, was the first of these viruses to be discovered when it was found to be a contaminant of polio vaccine (2). BK virus was identified from the urine of a kidney allograft recipient with chronic pyelonephritis and advanced renal failure in 1971 (3) and is the etiologic agent of polyomavirus-associated nephropathy. JCV was first identified in 1965 by electron microscopy in cases of progressive multifocal leukoencephalopathy (PML) (4,5) and was first isolated in culture in 1971 (6). Since then, JCV has been proven to be the opportunistic etiologic agent of this demyelinating disease. Recently, 2 new human polyomaviruses have been reported: the KI virus (7) and WU virus (8). However, the DNA sequences of these viruses indicate that they are more distantly related to SV40, BKV, and JCV than SV40, BKV, and JCV are to each other. These viruses also seem to have a distinct life cycle from SV40, BKV, and JCV (see succeeding sentences) in that they are associated with acute respiratory infections. Thus, they seem to represent a distinct new subfamily of human polyomaviruses, the characteristics of which are poorly understood at the present time.

Recently, a new human polyomavirus was reported to be associated with Merkel cell carcinoma (9). Southern blot evidence indicated that this virus is clonally integrated in tumor genomic DNA, suggesting that it has an early role in tumorigenesis. Future studies are needed to ascertain the natural history of this virus and its role in carcinogenesis.

In this brief review, we focus on JCV and, in particular, its potential role in the development of human cancer. This field can be subdivided into 3 main avenues of research. First, many studies have reported the potent oncogenic effects of JCV in experimental animals, including the production of solid tumors in rodents and monkeys and the development of neural tube origin tumors in transgenic mice. Second, molecular and cellular studies have pointed to the mechanisms whereby JC viral proteins such as T-antigen (T-Ag) and agnoprotein impact host cell functions that may be linked to oncogenesis. These include cell cycle control, alterations in DNA repair, and interactions with tumor suppressor proteins. Third, the presence of JCV DNA and the expression of JC viral proteins have been reported to be associated with tumors of the central nervous system (CNS) and, more recently, with non-CNS tumors. The animal studies and the molecular findings have been largely accepted, but the findings of the association of JCV with human tumors remain controversial. Possible technical and methodological differences that may contribute to the diverse findings are discussed.

The Normal Lifestyle of JCV: Latency and Reactivation

The human polyomavirus JC is a human neurotropic polyomavirus and is the opportunistic causative agent of PML, a fatal demyelinating disease of the brain that involves the cytolytic destruction of oligodendrocytes by replicating JCV. The high prevalence of antibodies in human sera against JCV indicates that JCV infection is widespread in the human population worldwide (10, 11). It is generally accepted that the virus infects most people during childhood and then remains in a persistent but dormant state known as latency, which is not well defined (reviewed in Refs. 12, 13). In most individuals, the level of JCV replication remains low, and infection is asymptomatic. However, in the context of severe immunosuppression such as AIDS, autoimmune diseases, agammaglobulinemia, lymphoma, and immunosuppressive drug treatment, JCV becomes reactivated in the CNS, and its replication in oligodendrocytes leads to the demyelinating disease PML (13, 14). Before the advent of AIDS in the early 1980s, PML was a very rare disease (15) but then increased dramatically in incidence (16). The high prevalence of PML in human immunodeficiency virus-1-infected patients makes it an AIDS-defining disease (17). Progressive multifocal leukoencephalopathy occurs in approximately 5% of AIDS patients (14, 18). In addition to its occurrence as an AIDS-associated disorder, PML was also recently observed in patients with multiple sclerosis undergoing integrin inhibitor therapy (reviewed in Refs. 19, 20) and in a lymphoma patient treated with the anti-CD20 monoclonal antibody rituximab (21).

Progressive multifocal leukoencephalopathy leads to a debilitating, most often fatal, neurologic disease (15, 18). Multifocal areas of demyelination in patients are usually detectable by magnetic resonance imaging. The 3 major histopathologic features of PML are foci of myelin loss in the subcortical white matter; enlarged oligodendrocytes with intranuclear eosinophilic inclusion bodies and bizarre astrocytes that are characterized by prominent pleomorphism; and atypical, lobulated, and frequently multiple nuclei (22, 23) (Fig. 1). Ultrastructural studies have demonstrated that the oligodendrocyte inclusions are the sites of active viral replication.


Progressive multifocal leukoencephalopathy. (A) Progressive multifocal leukoencephalopathy (PML) is characterized by multiple subcortical foci of demyelination highlighted using Luxol fast blue stain for myelin (original magnification: ×20). (B) Higher magnification of several demyelinated lesions (hematoxylin and eosin; original magnification: ×100). (C) Bizarre astrocytes with extreme pleomorphism and multiple lobulated, atypical nuclei are characteristic cells in PML. (D) Immunohistochemistry for VP1 in a bizarre astrocyte. (E) Enlarged oligodendrocytes with intranuclear eosinophilic inclusion bodies are pathognomonic for PML (hematoxylin and eosin stain). (F) The inclusions are the site of active viral replication as demonstrated by expression of the capsid protein VP1. (D-F), original magnification: ×1,000. (G) Characteristic icosahedral JC viral particles are shown by electron microscopy in the nucleus of an infected oligodendrocyte in a case of PML.

The transmission of JCV within the human population and the life cycle of JCV within the body are poorly understood. The transmissible form of JCV is known as the archetype (JCVCY), which is excreted in the urine and is found in sewage (24, 25). The pathologic forms of JCV that have been isolated from the brains of patients with PML are characterized by the occurrence of rearrangements, including deletions, duplications, and point mutations in a specific region of the JCV control region (NCCR); these are designated as "PML-type" JCV, for example, Mad-1 and Mad-4. The archetype JCV is found in the kidney of normal individuals and is thought to replicate episodically or at low levels, probably under conditions of low immune potency, to give rise to virus that is shed in the urine.

The nature of the latent state of JCV is not well characterized but is thought to involve an asymptomatic, chronic, persistent infection where JCV DNA can be detected by highly sensitive polymerase chain reaction (PCR) techniques, but expression of JCV proteins cannot be detected by immunohistochemistry or Western blot. Many tissues have been reported to harbor latent JC virus, including the kidney (26), tonsils (27), gastrointestinal (GI) tract (28, 29), and the brain (30-33). Interestingly, whereas the kidney harbors the archetype transmissible strain of JCV that can be shed intermittently into the urine (25, 26), it has been reported that latent JCV found in the normal brain contains rearrangements in the transcriptional control region (NCCR), that is, it is PML-type JCV (30). Human polyomavirus JC is able to infect B-lymphocytes (27), and it is possible that JCV can be transported from organs such as the tonsils and the kidney into the CNS by infected blood cells, at which time rearrangement of the NCCR may occur. The dissemination of JCV with mutated NCCR in the circulation may have a role in the development of JCV-associated tumors and is discussed in the succeeding sentences. Indeed, we have reproduced and confirmed the previous report (31) of PML-type JCV DNA in a series of normal brain clinical samples (in our case, Mad-4) and have further shown by laser capture microdissection that this DNA sequence is present in oligodendrocytes and astrocytes (the cell of origin of oligodendrogliomas and astrocytomas, respectively) but not in neurons (34). This interesting observation requires further investigation.

JCV and Cancer: In Vitro and Experimental Animal Studies

Human polyomavirus JC has the capacity to transform cells such as human fetal glial cells and primary hamster brain cells in vitro. Human polyomavirus JC-transformed cells exhibit a typical "transformed phenotype," that is, growth in soft agar, serum independence, morphologic changes, multinucleation, plasminogen activator production, and enhanced glucose uptake (35). The transforming ability of JCV seems to be limited to cells of neural origin possibly because the transcriptional regulation of the viral promoters (NCCR) involves cell-type-specific transcription factors.

Many studies have also established the highly oncogenic potential of JCV in laboratory animals. For example, JCV was first found to induce multiple types of brain tumors when it was injected into the brains of newborn Golden Syrian hamsters (36). Human polyomavirus JC is the only human virus that induces solid tumors in nonhuman primates. It caused the development of astrocytomas, glioblastomas, and neuroblastomas in owl and squirrel monkeys 16 to 24 months after they were inoculated with the virus intracerebrally, subcutaneously, or intravenously (37, 38). Studies of the monkey tissue revealed T-Ag expression but no capsid protein or infectious virions, indicating that monkey cells are nonpermissive for JCV replication (37-39). Interestingly, JC viral DNA was integrated into the cellular DNA of JCV-induced tumors that arose in owl monkeys (40). These animal studies have been reviewed recently (35, 41, 42). The features of the experimental tumors may be relevant to the JCV-associated human cancers discussed in the succeeding sentences, particularly with respect to the nonpermissive nature of the JCV infection, the high level of T-Ag expression, the absence of any late capsid protein expression or viral replication, and the occurrence of alterations to the viral DNA.

Transgenic mice containing only the JCV early region expressing T-Ag can develop tumors of the CNS. Small et al (43) reported in 1986 the development of adrenal neuroblastomas, and Franks et al (44) generated a line of transgenic mice with the JCV T-Ag gene under the control of the natural JCV promoter and exhibit tumors of primitive neuroectodermal origin. Transgenic mice with the JCV T-Ag gene can also develop tumors that arise from the pituitary gland (45). Moreover, it has also been reported that a small subset of these animals developed solid masses arising from the soft tissues surrounding the salivary gland, the sciatic nerve, and along the extremities that are histologically similar to malignant peripheral nerve sheath tumors (46). Thus, transgenic mice expressing the JCV early region develop tumors of both CNS and non-CNS tissues. The types of tumor formed seem to be influenced by the nature of the JCV NCCR and the mouse strain used (35, 41, 42). Recently, we have studied the frequency of the different types of tumors in mice that are transgenic for the archetype, JCVCY strain early region. These animals developed tumors of neural crest origin, including primitive neuroectodermal tumors, adrenal neuroblastomas, pituitary tumors, glioblastoma multiforme, and malignant peripheral nerve sheath tumors. Of these, primitive neuroectodermal tumors are the most common and arise in all parts of the neuroaxis, including the brainstem, basal ganglia, and hemispheres, in order of frequency. Neoplastic cells in all of the different tumor types express the JC early product T-Ag, which colocalizes with the cell cycle protein p53 in the nuclear compartment, but normal tissues that also contain the transgene show no expression of T-Ag, suggesting specificity of the T-Ag in the oncogenic process. The many similarities between the tumors that develop in the JCV transgenic animals and the human tumors of neuroectodermal origin make this animal model an excellent tool for the study of oncogenesis in humans (unpublished observation). Figure 2 illustrates representative observations examples of T-Ag-expressing tumors developed by the JCVCY transgenic mice.


Human polyomavirus JC-induced tumors in transgenic animals. (A) The brain of a JCVCY transgenic mouse with a large, extra-axial mass in the pituitary region. (B) Consecutive sections of the brain of a mouse with an intra-axial primitive neuroectodermal tumor centered in the brainstem and infiltrating into the basal ganglia (hematoxylin and eosin [H&E] stain; left panel). Immunohistochemistry for T-antigen (T-Ag; right panel) demonstrates protein expression by the tumor but not the adjacent brain. (C) Primitive neuroectodermal tumors are characterized by sheaths of closely packed neoplastic cells with slightly elongated nuclei (H&E; left panel). The tumor cells express cytoplasmic Class III β-tubulin (middle panel) and T-Ag in the nuclei (right panel). (D) A pituitary tumor is composed of round neoplastic cells with abundant cytoplasm (H&E; left panel) and expresses cytoplasmic prolactin (middle panel) and nuclear T-Ag (right panel) by immunohistochemistry. (E) Malignant peripheral nerve sheath tumor is composed of fascicles of spindle-shaped cells with elongated nuclei (H&E; left panel). These tumors express S-100 (middle panel) and nuclear T-Ag (right panel) by immunohistochemistry. Original magnification for all histopathological panels: ×1,000.

JCV and Cancer: Studies on the Molecular Mechanisms of Action of JC Viral Proteins

In addition to the viral capsid structural proteins (VP1, VP2, and VP3), JCV expresses 3 proteins with regulatory functions: 1) the large T-Ag, 2) the small t-antigen encoded in the early region, and 3) agnoprotein encoded in the late region. Three splice variants of T-Ag have been reported (47). In addition to regulating the JC viral life cycle, these proteins are involved in interactions with host proteins and dysregulating the cellular processes that lead to changes that may be involved in malignant transformation.

Human polyomavirus JC T-Ag is a nuclear phosphoprotein of 688 amino acids. It is a very versatile protein in that it interacts with the polyomavirus DNA at the viral origin of replication and initiates viral DNA replication; it also interacts with a plethora of cellular proteins that are responsible for driving cells into S-phase. Therefore, it is a major factor in cellular transformation. Early studies had demonstrated that SV40 T-Ag interferes with 2 key tumor suppressor proteins that regulate cell cycle progression, pRb and p53. This aberrant stimulation of the cell cycle is a driving force for oncogenic transformation (reviewed in Refs. 13, 41, 42, 48). Human polyomavirus JC T-Ag also binds and regulates these 2 key signaling molecules (49). The interaction of T-Ag with pRb is thought to activate E2F transcription factors that promote cell cycle progression (50), whereas interaction with p53 is thought to compromise its protective role against both DNA damage and oncogenic transformation.

Other studies also indicate that T-Ag can modulate signaling proteins in addition to pRb and p53. Human polyomavirus JC T-Ag directly binds to insulin receptor substrate 1 (IRS-1) and causes its translocation to the nucleus (51); there is growing evidence that the interaction of JCV T-Ag with IRS-1 contributes to the process of malignant transformation in medulloblastomas, the most prevalent solid brain tumor of childhood (52). Moreover, the interaction of T-Ag with IRS-1 impairs homology-directed DNA repair (53). Human polyomavirus JC T-Ag also directly binds to β-catenin (54, 55). This interaction causes β-catenin to translocate to the nucleus, where it enhances expression of genes such as c-myc and cyclin D1. Alternatively, a noncanonical pathway that allows T-Ag to recruit Rac1 for stabilizing β-catenin by inhibiting its ubiquitin-dependent proteasomal degradation has been identified (56).

In mice transgenic for the JCV early region, JCV T-Ag associated with the neurofibromin 2 protein (also known as merlin); the tumors that developed were histologically characterized as malignant peripheral nerve sheath tumors in which neurofibromin 2 was present in a ternary complex with T-Ag and p53 (46). Current available data suggest that neurofibromin 2 is a positive regulator of p53 in tumor suppressor activity, and that T-Ag may inhibit this activity.

The second of the 2 proteins encoded by the early region of primate polyomaviruses is the 172-amino acid small t-antigen. The N-terminus of small t-antigen is shared with the N-terminus of T-Ag, but the C-terminus is a unique domain that is incorporated by alternative splicing of the early region primary transcript. Research on primate polyomavirus small-t has concentrated mainly on SV40, and it seems to function primarily by inhibiting the cellular phosphatase PP2A, thereby activating kinase signaling pathways that control cell proliferation (13, 41, 42, 48).

Finally, the late region of JCV encodes the 71amino acid, highly basic agnoprotein. Agnoprotein expression has been detected by Western blot during infection of glial cells by JCV (57-60). In infected cells, the 8-kd agnoprotein is produced late in the viral cycle and is found mainly in the cytoplasm, especially in the perinuclear region; a small amount may also be found in the nucleus (61). Agnoprotein expression can exert profound effects on the cells, including dysregulation of cell cycle progression and cell accumulation at the G2/M phase (62). A decline in cyclin A- and cyclin B-associated kinase activity was also observed in these cells. Agnoprotein increased the activity of the p21/WAF-1 promoter and the level of p21/WAF-1 protein; it was shown to bind directly to p53. Activation of p21/WAF-1 gene expression in agnoprotein-expressing cells is thought to be mediated, at least in part, through its cooperation with p53 (62). Ectopic agnoprotein expression also affects the response of cells to DNA damage. Agnoprotein-expressing cells were more sensitive to the cytotoxic effects of cisplatin and exhibited increased chromosome fragmentation and micronuclei formation (63). This seems to be due to agnoprotein binding to the Ku70 DNA repair protein and inactivating it by sequestering it in the perinuclear space. Indeed, agnoprotein synthesized in vitro can directly bind to Ku70 in cell extracts and inhibit nonhomologous end-joining-mediated DNA repair. Agnoprotein also impaired DNA damage-induced cell cycle arrest (63).

The effects of agnoprotein and T-Ag on DNA repair and the cellular response to DNA damage raise the issue of the importance of DNA damage in the oncogenic functions of JCV. In addition to their ability to transform cells by virtue of the interactions of the viral proteins with cellular signaling proteins, there is considerable evidence that the polyomaviruses in general have a mutagenic effect on cellular DNA. It is therefore possible that secondary mutations induced by JCV contribute to the pathogenesis of JCV-associated tumors. This literature has been reviewed elsewhere (48). In this regard, we recently analyzed in detail the effects of JCV infection of primary human glial cells on the cellular processes involved in DNA damage and repair (64). Genomic stability and DNA repair were found to be significantly dysregulated by JCV infection. Metaphase spreads exhibited increased ploidy correlating with duration of infection. Increased micronuclei formation and phospho-histone2AX expression also indicated DNA damage. Western blot analysis revealed perturbation in expression of some DNA repair proteins, including a large elevation in the level of Rad51. This elevation was also found to occur upon JCV infection in vivo because immunohistochemistry on clinical samples of PML showed robust labeling for Rad51 in the nuclei of bizarre astrocytes and inclusion body-bearing oligodendrocytes that are characteristic of JCV infection. Finally, it should be noted that the effects of acute JCV infection on Rad51 might differ significantly to the effects of chronic T-Ag in tumor cells in which T-Ag attenuates faithful DNA repair by forcing nuclear interaction between IRS-1 and Rad51 (51-53). The inhibition of homologous recombination may force T-Ag-expressing cells to repair DNA double-strand breaks by using nonhomologous end-joining, an alternative mechanism that does not require a DNA template but is considered error-prone and may lead to the accumulation of mutations.

Recent studies suggest that the complex series of events unleashed by T-Ag expression may involve additional pathways. For example, apoptosis is a cellular homeostatic mechanism to dispose of damaged or virus-infected cells, and despite the extensive DNA damage caused by JCV infection, there is no evidence for apoptosis in infected cells or in cases of PML. This may be at least in part because of a novel discovery. Human polyomavirus JC infection and T-Ag expression in particular induce the expression of an anti-apoptotic protein, survivin. This member of the inhibitors of apoptosis family is normally expressed during embryonic development, but its expression should be silenced in fully differentiated and mature tissues (65). T-Antigen has the ability to bind and activate the survivin promoter, resulting in a significant decrease in apoptosis, which is dramatically increased after survivin siRNA treatment. We hypothesize that JCV activates this pathway to have enough time to replicate and complete its lytic cycle (66). The implications for astrocytes that do not undergo lytic infection are problematic in the long term because the inhibition of apoptosis may contribute to prolonged life span, accumulation of additional mutations, and the transformed phenotype.

Taken together, current data indicate that in addition to its ability to transform cells by disrupting cellular signaling pathways, JCV also has a mutagenic effect on cellular DNA and induces karyotypic instability through the actions of T-Ag and agnoprotein. It is therefore possible that secondary mutations induced by JCV contribute to the development of neoplasia.

JCV and Cancer: Studies on Clinical Samples From Human Tumors That Indicate a Positive Association

There are more than 25 reports of the association of JCV with human cancer either by virtue of the occurrence of the cancer concomitantly with PML or by the molecular detection of viral footprints in neoplastic cells. Because we have recently published a comprehensive review of these studies (48), they are summarized briefly here.

The first indication for the association of JCV with brain tumors came from reports of brain tumors found in patients with concomitant PML such as the case reported by Richardson (67), who first described PML. In that case, a patient with chronic lymphocytic leukemia and PML was found to have an oligodendroglioma on postmortem examination. Other reported cases have reviewed (35). The finding that brain tumors are associated with PML is consistent with a role for JCV in the development of these types of tumors, but PML might develop secondarily to the immunosuppression that can be associated with the cancer.

Human polyomavirus JC has also been reported to be associated with brain tumors in patients without PML. In 1996, Rencic et al (68) reported JCV DNA detected by PCR in tumor tissue from a patient with an oligoastrocytoma. The identity of the amplified PCR product was confirmed as JCV by DNA sequencing. Moreover, JCV RNA and T-Ag protein were detectable in the tumor tissue by primer extension analysis and Western blotting, respectively. This study indicated that JCV gene expression occurred in the tumor cells. Del Valle et al (69) examined 85 samples of various glial tumors for the presence of JCV DNA sequences and T-Ag expression. They found that, depending on the tumor type, 57% to 83% of tumors were positive for JCV. In other reports, JCV has been associated with CNS lymphoma, glioblastoma multiforme, oligoastrocytoma, oligodendroglioma, medulloblastoma, and xanthoastrocytoma (35, 48). Examples of JCV T-Ag expression in representative human brain tumors are illustrated in Figure 3.


Detection of T-antigen (T-Ag) in human brain tumors. Immunohistochemistry for T-Ag (right side panel in each pair) in a series of brain tumors. Astrocytic tumors ranging from diffuse fibrillary astrocytoma to glioblastoma multiforme (upper half of figure with immunolabeling for glial fibrillary acidic protein in the left panel in each pair) exhibit nuclear expression of the oncoprotein in neoplastic cells. T-Antigen is also expressed in the nuclei of tumor cells in an oligodendroglioma and an ependymoma (typical features shown in hematoxylin and eosin stain in left panels of each pair). A pathognomonic rosette is shown in the ependymoma. T-Antigen is also detected in a medulloblastoma that expresses Class III β-tubulin and in a primary CNS lymphoma that expresses CD-20 in a perivascular cuff of neoplastic cells (left side panels in the bottom row pairs).

In addition to CNS tumors, it has also been found that JCV may have an important role in tumors outside of the CNS. Human polyomavirus JC genomic sequences have been detected in normal tissue samples taken from the upper and lower human GI tract and in colon cancer (40, 70). In a study of malignant epithelial tumors of the large intestine, 22 of 27 samples contained JCV genomic sequences amplified by PCR. Expression of JCV T-Ag and agnoprotein was observed in greater than 50% of the samples, whereas no expression of JCV proteins was detected in normal GI epithelial tissue (54). Similar results have been found for the upper GI tract. JC viral DNA was isolated from 11 (85%) of 13 normal esophageal biopsies and from 5 (100%) of 5 esophageal carcinomas. By immunohistochemistry, JCV T-Ag was detected in 10 (53%) of 19 carcinomas, agnoprotein in 8 (42%), p53 tumor suppressor in 11 (58%), and β-catenin in 4 (21%). None of 51 normal, benign, or premalignant esophageal samples expressed viral proteins. Laser capture microdissection verified the presence and specificity of JCV DNA sequences. β-Catenin and p53 colocalized with JCV T-Ag in the nuclei of neoplastic cells (28). In a study of gastric tissue samples, 21 (57%) of 37 gastric cancers harbored JCV T-Ag sequences, and 13 (30%) of 37 gastric cancers contained VP-1 DNA sequences. T-Antigen protein expression was found in 9 (39%) of 23 gastric cancers, whereas no expression was observed in any of the nonneoplastic tissues (71). A recent study of a panel of lung tumors using in situ PCR and immunolabeling revealed JCV positivity in the nuclei of lung carcinoma cells, and a role for JCV has been proposed for in-lung carcinogenesis especially in tumor types other than adenocarcinoma (72). Another recent study has also reported the presence of JCV DNA and expression of T-Ag in human lung tumors (73).

JCV and Human Cancer: Studies That Fail to Detect a JCV Association With Human Tumors

Despite the mounting evidence pointing to an association of JCV with human cancer, there have been some reports that have failed to detect JCV DNA sequences in panels of human tumors. These reports and some technical comments that may be of interest are discussed below.

Herbarth et al (74) analyzed 30 brain tumors classified as oligodendroglioma or astrocytomas and 22 human glioma cell lines by PCR and failed to amplify JCV DNA using 2 different primer sets to opposite sides of the genome. Furthermore, T-Ag expression was not detectable by Northern blot of RNA isolated from the tumor. Controls for sensitivity were included and consisted of amplification of JCV sequences from plasmid DNA in serial dilutions. Dilution of plasmid DNA with salmon sperm was also shown not to inhibit amplification. However, a number of technical points should be noted. First, DNA isolated from formalin-fixed paraffin-embedded tumors is usually of inferior quality (75), so that comparison with high-quality DNA isolated from plasmid or from tissue culture cells may not be an appropriate control. The authors use ethidium bromide staining to identify DNA bands. This is important in light of previous studies showing that only a few tumors have sufficient JCV copy number to give such a band, and a more sensitive method using Southern blot was required for detection of JCV DNA in more tumors (35, 68, 69, 76, 77). Second, only a subpopulation of tumor cells may express T-Ag, and, thus, Northern blot with 10 μg of total cellular RNA is unlikely to detect T-Ag mRNA even if the RNA is perfectly preserved. The positive control glyceraldehyde-3-phosphate dehydrogenase RNA is highly abundant in all cells. A much better and more sensitive method would be to perform immunohistochemistry and laser capture microdissection followed by PCR (see succeeding sentences).

In a similar study, Weggen et al (78) analyzed 30 brain tumors showing JCV sequences in only 1 of 131 meningioma samples. The previously discussed technical issues regarding the comparison of DNA from formalin-preserved samples to plasmid DNA and the lack of Southern blot may also apply to this study, which also examined the presence of SV40 and BKV with largely negative results. In this study, immunohistochemistry for T-Ag and VP1 was performed with negative results. It should be noted, however, that the immunohistochemistry was performed with only 1 meningioma sample that was SV40 positive, so the significance of this finding is unclear. Similarly, Arthur et al (79) reported lack of detection of BKV or JCV by PCR from primary glioblastoma multiforme using DNA extracted from 33 fresh-frozen tumors and paraffin-embedded tumors. A more recent study by Muñoz-Mármol et al (80), which used a similar approach to that used in our laboratory, nevertheless only detected 5 PCR-positive samples of 55 gliomas, 5 medulloblastomas, and 15 reactive gliosis cases. That study used paraffin-embedded tissue samples only. Immunodetection of T-Ag in human brain tumor panels has also yielded some negative reports. Sabatier et al (81) recently reported that none of a French series of 82 CNS tumors was positive for T-Ag by immunohistochemistry. This study used the same monoclonal antibody (Ab-2, clone Pab416; Oncogene Research, San Diego, CA) that was routinely used in immunohistochemistry of brain tumors with positive results (29, 68, 69, 76).

Recently, it was reported that polyomavirus DNA was rarely present in a series of brain tumor tissue specimens examined by PCR, followed by Southern hybridization (82). This discrepancy is most likely due to methodologic issues related to the efficiency of the DNA extraction, the sensitivity of either the PCR or the Southern blot, the copy number of positive control plasmid, problems with paraffin-embedded archived tissue, and other issues that we have discussed in detail (83). The unambiguous detection of T-Ag in the nuclei of neoplastic cells and JCV agnoprotein in perinuclear regions are findings that could not result from any type of laboratory contamination (69, 83-85). Furthermore, JCV DNA has been isolated from brain tumors and analyzed by DNA sequencing. These DNA sequences sometimes contain mutations when compared with established strains of JCV. For example, in JCV DNA detected in 5 cases of oligodendroglioma, 3 contained the Mad-4 control region (NCCR), and 2 contained the JCVCY configuration with a number of novel mutations (86). In another study, we investigated a patient with an unusual case of glioblastoma multiforme composed of 2 components, 1 with the characteristics of glioblastoma multiforme and the other being a small cell neuronal-like component. We isolated and sequenced JCV DNA from both tumor compartments and found that both contained the Mad-1 form of the NCCR but with multiple mutations throughout the sequence (76). Interestingly, the point mutations that were found in the 2 compartments of this tumor were different, suggesting that either dual viral infections occurred, or, more likely, that divergent mutations occurred during the development of this very unusual tumor. In these sequencing studies, template DNA was isolated directly from tumor samples without an intermediate cloning step in bacteria. This excludes the possibility that the mutations are due to cloning artefacts.

There is also controversy concerning a possible role for JCV in colorectal tumors. As described in the previous section, several groups have reported an association of JCV with these tumors. In contrast, Newcomb et al (87) analyzed 45 urine samples from normal donors and 233 colorectal cancer/normal tissue pairs and found only 1 tissue sample (a normal tissue) to be positive while detecting JCV in 70% of the urine samples. In a follow-up letter, it was argued that great care is needed to optimize PCR to detect small circular DNA present in low amounts in clinical samples, and that the use of formalin-fixed paraffin-embedded tissue can be problematic (88). In response to these criticisms, the authors countered that they can amplify a single-copy gene from their DNA samples from formalin-fixed, paraffin-embedded tumors, and that they insist on very stringent criteria for calling a sample "JC virus-positive," and that sequence variations in JCV isolates may represent errors introduced by Taq polymerase, rather than natural viral sequence variants (88). These issues reflect the problems that arise in this field.

It remains a puzzle as to why studies seeking to identify the presence of JC viral DNA in normal and tumor tissues have yielded such variable detection rates. The rate of detection in normal tissues can be quite high or zero, and similarly, tumors can range from high to none. It seems likely that technical issues related to PCR and sample preparation are at play as we have mentioned. False-negative results may occur if PCR sensitivity is low or sample DNA is degraded. Equally, because JCV infection occurs in 70% to 80% of people, and JCV can infect blood B-lymphocytes, (which would be expected to be present in tissue samples), false-positive results may occur. In this regard, immunohistochemical labeling followed by laser capture microdissection of specific cell subpopulations has allowed the PCR amplification and sequence analysis of JCV associated with specific cell types, for example, T-Ag-positive cells within a tumor (28, 76), providing stronger evidence for the association of JCV.

The extent of association of JCV with human tumors is still an area of debate. New developments in technology to address the questions in new ways would be welcome. More collaboration between laboratories and exchanges of samples would also help advance the field. Because a role of JCV in tumorigenesis might have therapeutic implications, it is important that resolution of these issues be given high priority.

Additional Studies and Key Questions

Associations of certain DNA tumor viruses with human cancers are well established. Epstein-Barr virus is associated with Burkitt lymphoma (77), and human papilloma virus is associated with cervical carcinoma (89). Indeed, human papilloma virus infection is a necessary factor in the development of nearly all cases of cervical cancer. Because JCV is such a common virus, however, it is difficult to establish linkage to causation, rather than association of JCV with neoplasms. The carcinogenesis of human papilloma virus and cervical cancer is associated with the expression of the viral early oncogenic proteins E6 and E7, which bind and inactivate p53 and pRb, to induce a continuous cell proliferation with the increasing risk of accumulation of DNA damage that eventually leads to cancer (90). Similarly, JCV exerts a variety of molecular manipulations, including the action of the oncogenic protein large T-Ag, in the binding and inactivation of p53 and pRb and induction of DNA damage (see previous sentences). However, such mechanisms require that JCV be present at 1 or more copies per cell, and in tumors where JCV may be present at less than 1 copy, it is not possible for every cell to express T-Ag, and other mechanisms may be involved in the mutagenic activity of JCV.

Although JCV is a potent tumor virus in experimental animals, its role in human cancer is not understood, and no consensus has been reached on the extent of its involvement. Human polyomavirus JC is a ubiquitous virus, being present in greater than 80% of people, making it difficult to perform epidemiologic studies. In this regard, a study of space-time clustering of childhood brain tumors (especially astrocytoma) has suggested the existence of an infectious etiologic agent (91, 92). As previously discussed, it has not yet been possible to draw conclusions regarding the relative abundance of JCV DNA in JCV-associated tumors compared with JCV DNA present in normal tissue. In addition, whereas JCV DNA may be detectable in normal tissues, there is no expression of any viral protein as judged by immunohistochemistry for T-Ag, agnoprotein, or VP1. In contrast, expression of nuclear T-Ag and perinuclear agnoprotein, but not VP1, is clearly visible in immunohistochemistry of JCV-associated tumors (69, 84, 85).

What is the nature of JC viral DNA in tumors? Human polyomavirus JC may be carried around the blood stream in B-lymphocytes, and viral DNA rearrangements may occur during this passage. Although this may have a role in the genesis of rearranged JCV genomes (PML-type JCV configurations) in PML, a role of these processes for JCV-associated tumors is speculative at this time. Interestingly, some JCV-associated tumors contain viral DNA with PML-type sequences, suggesting that the rearrangement mechanisms that alter the viral NCCR DNA during the pathogenesis of PML may also occur for virus that becomes associated with tumors. The recently discovered human Merkel cell polyomavirus is present as a clonally integrated insert in the genomic DNA of Merkel cell carcinoma tumors, suggesting that viral infection and integration is an early event in tumorigenesis (9), but no such evidence exists for JCV.

Whereas many tumors contain the complete JCV genome, some tumors contain subgenomic fragments of JCV DNA as shown by parallel Southern blot analyses with probes from different parts of the JCV genome. This suggests either that cells can take up fragmented JCV DNA or that this DNA represents the footprint of an earlier abortive DNA infection, which was followed by the partial elimination of the viral DNA. In addition, sequence analysis of the NCCR of JCV DNA isolated from these tumors can also harbor novel point mutations. For example, for JCV DNA from 5 oligodendroglioma, 3 contained Mad-4 sequence, and 2 contained the JCVCY configuration with a number of novel mutations (86). In another study of a patient with glioblastoma multiforme, we isolated JCV DNA with the Mad-1 form of the NCCR but with multiple mutations throughout the sequence (76). In that study, the glioblastoma multiforme was associated with a small cell neuronal cell component that was histologically distinct from the rest of the tumor, and JCV DNA from that tumor component had a different set of point mutations from the JCV DNA from the main tumor. This suggests that either 2 distinct JCV infection events occurred or that differential evolution of the JCV DNA in the 2 cell lineages occurred after 1 JCV infection event (76).

It is important to stress that, although several studies have found that JCV VP1 is not expressed in JCV-associated tumors, indicating that viral replication is not occurring, expression of JCV T-Ag and agnoprotein is usually robust. T-Antigen expression was found to occur in some, but not all, of the tumor cells containing JCV DNA in JCV-associated human medulloblastoma (93, 94). This is also the case in JCV-transgenic mice in which immunohistochemical analysis of medulloblastoma-such as cerebellar neuroectodermal origin tumors showed that T-Ag expression was not seen in all of the tumor cells, that is, the tumor comprised a heterogeneous population of T-positive and T-negative cells (95). Indeed, we were able to clone out these T-positive and T-negative cells, and we found that the T-negative cells contained a mutant truncated form of p53 that had lost the transcriptional transactivation activity of wild-type p53 (96). This suggests that during the course of tumor evolution of the JCV T-Ag mouse medulloblastoma, a mutation occurred that inactivated p53, allowing tumor progression even in the absence of continued T-Ag expression. This may represent an example of a way in which JCV T-Ag can induce host DNA damage, inactivate tumor suppressors, and contribute to tumor evolution early in tumorigenesis. In this scenario, we speculate that an abortive JCV infection event may be involved in tumor initiation involving T-Ag expression, which is then followed by loss of T-Ag expression as the tumor evolves.

A second scenario can be envisioned in which JCV infection may occur after tumor initiation. Although JCV can be found in normal tissue, it is latent, and gene expression cannot be detected. However, it is possible that infection of preneoplastic cells, which may have altered profiles of cellular transcription factors, allows expression of the early region leading to T-Ag production and tumor progression. In this scenario, T-Ag may act as a tumor promoter or accelerator, rather than an initiator. The 2 scenarios are not mutually exclusive.

Another important question is why early T-Ag expression is observed in JCV-associated tumors but not late capsid expression and why the virus does not replicate. Gene expression and DNA replication in polyomaviruses are restricted by cellular factors that are tissue- and species-specific (97). During the course of productive infection of permissive cells by polyomaviruses, a transcriptional switch from expression of the early proteins to expression of the late proteins occurs. This has been well studied in the model virus SV40, where T-Ag has been shown to repress early mRNA production, that is, autoregulation (98-100), and to stimulate late mRNA production (101). Similarly, JCV T-Ag has been shown to stimulate transcription of the JCV late promoter (102,103). In tumors induced by JCV in experimental animals, tumors arise in the brain likely because of a requirement of the JCV early promoter for glial cell tissue-specific factors to allow T-Ag expression, whereas the lack of late gene expression and viral replication in response to T-Ag expression is probably due to a requirement for human species-specific cellular factors. In human tumors associated with JCV, the reason for the observed absence of late gene expression and viral replication is not clear. Interestingly, JCV-positive tumors typically show more robust nuclear immunolabeling for T-Ag than cells from PML clinical samples in which virus is actively replicating (Del Valle, unpublished observations). Because T-Ag is abundant and human species-specific transcription factors are presumably present, it would be expected that the late promoter would be activated. There are a number of possibilities that might be considered for why this does not occur. First, the virus may have acquired mutations in the control region or deletions in the late structural genes that prevent capsid expression. Another possibility is that the virus infected a preneoplastic cell that may have altered profiles of cellular transcription factors, allowing the expression of the early region but not the late region. Alternatively, unknown stochastic epigenetic events might prevent late gene expression.

Summary and Conclusions

Overall, the current available data suggest that JCV is a ubiquitous virus that usually exists in a latent state characterized as an asymptomatic, chronic, persistent infection but where the virus is able to spread by intermittent and subclinical infection events that are nonetheless productive but tightly controlled by the immune system. Under conditions of severe immunosuppression, lytic and pathologic infection of glial cells can occur, causing PML. It is possible that JCV involvement in tumors may be caused by very rare infection events that are abortive and nonproductive due to a failure to express the viral late genes, giving rise to clades of cells that are driven to grow progressively by their expression of T-Ag but do not undergo viral-induced cell lysis. The cellular signaling pathways that have been identified as targets of JCV T-Ag in molecular experiments and in experiments with JC early region transgenic mice are the same pathways that are observed to be dysregulated in human tumors that are immunopositive for T-Ag, that is, the p53 tumor suppressor protein (49, 69, 96), the retinoblastoma protein pRb and its family members (49, 50), the insulin-like growth factor I/IRS-1 signaling system (51, 52), and the Wnt/β-catenin signaling pathway (54-56). By their nature, early events in the genesis of neural tumors are difficult to study, but JCV involvement is a possibility that can neither be implicated nor ruled out at the present time.


The authors thank past and present members of the Department of Neuroscience and Center for Neurovirology for continued support, insightful discussion, and sharing of reagents and ideas and C. Schriver for editorial assistance.


  • This work was made possible by grants awarded by the National Institutes of Health to L.D.V., M.K.W., and K.K.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
View Abstract