Human neurotropic polyomavirus, JCV, and its role in carcinogenesis

Kamel Khalili*,1, Luis Del Valle1, Jessica Otte1, Michael Weaver1, Jennifer Gordon1

1Center for Neurovirology and Cancer Biology, College of Science and Technology, Temple University, 1900 North 12th Street,015-96, Room 203, Philadelphia, PA 19122, USA

A number of recent studies have reported the detection ofthe ubiquitous human polyomavirus, JC virus (JCV), insamples derived from several types of neural as well asnon-neural human tumors. The human neurotropic JCVwas first identified as the etiologic agent of the fataldemyelinating disease, progressive multifocal leukoence-phalopathy, which usually occurs in individuals withdefects in cell-mediated immunity, including AIDS.However, upon mounting evidence of the oncogenicpotential of the viral regulatory protein, T-antigen, andJCV’s oncogenecity in a broad range of animal models,studies were initiated to determine its potential involve-ment in human carcinogenesis. Initially, the mostfrequently observed tumors in rodent models, includingmedulloblastoma, astrocytoma, glioblastoma, and otherneural-origin tumors were analysed. These studies werefollowed by analysis of non-neural tumors such ascolorectal carcinomas. In a subset of each tumor typeexamined, JC viral genomic DNA sequences could bedetected by PCR and confirmed by Southern blothybridization or direct sequencing. In a smaller subset ofthe tumors, the expression of T-antigen was observed byimmunohistochemical analysis. Owing to the establishedfunctions of T-antigen including its ability to interact withtumor suppressor proteins such as Rb and p53, and itsability to influence chromosomal stability, potentialmechanisms of JCV T-antigen-mediated cellular dysre-gulation are discussed. Further, as increasing evidencesuggests that T-antigen is not required for maintenance ofa transformed phenotype, a hit-and-run model forT-antigen-induced transformation is proposed.Oncogene (2003) 22, 5181–5191. doi:10.1038/sj.onc. 1206559

Keywords: PML, oncogene, cancer, brain

Introduction

The human neurotropic JC virus (JCV) is a member ofthe polyomavirus family, which includes the human BKvirus (BKV) and the simian virus 40 (SV40). Both JCVand BKV are thought to infect approximately 80% ofthe human population in early childhood and toestablish latency in the kidney. Upon reactivationbecause of immunosuppression, JCV induces the once

rare demyelinating disease progressive multifocalleukoencephalopathy (PML) most frequently seen inAIDS patients, while BKV induces polyomavirusnephropathy, an increasingly common side effect ofimmunosuppressive therapy in renal transplant recipi-ents. While the question as to whether SV40 represents ahuman polyomavirus, or was introduced into humansthrough contaminated poliovaccine is still under debate,serologic evidence suggests that SV40 now circulateswithin the human population. Although a potential siteof latency for SV40 has not been established, a numberof studies have reported the detection of the virus inhuman tumors including mesotheliomas and osteosar-comas, as well as two rare CNS tumors, ependymomasand choroid plexus papillomas (reviewed in Arringtonand Butel, 2001). In addition, there have been severalreports of BKV DNA detected in normal tissues and celllines including the urogenital tract, brain, bone,peripheral blood cells, as well as one report in whichBKV RNA was detected in bone, brain, and peripheralblood cells by Northern blot or RT–PCR raising thequestion as to whether polyomaviruses may be presentin normal tissues or peripheral blood (reviewed in DeMattei et al., 1995; Corallini et al., 2001). In this review,we will consider the literature concerning the oncogenicpotential of JCV in experimental animal models andsummarize what has been reported regarding theassociation of JCV with human neoplasias. Theseobservations, coupled with in vitro experiments aimedat understanding the molecular pathways affected byviral gene expression, will help elucidate potentialmechanisms of tumorigenesis by JCV.

Progressive multifocal leukoencephalopathy

First described in 1958, PML was once considered a raredisorder, which occurred in individuals with immuno-suppressive disorders such as Wiscott Aldridge syn-drome, individuals receiving chemotherapy, orinfrequently in elderly individuals (Walker and Padgett,1983; Price, 1996). Since the AIDS epidemic, however,PML has become an increasingly common neurologicalcomplication in the developed world in that up to 10%of HIV-1-positive individuals are predicted to becomesymptomatic during the course of their antiretroviraltherapy (Berger and Concha, 1995). Individuals suffer-ing from PML usually present with decreased motor*Correspondence: K Khalili; E-mail: kamel.khalili@temple.edu

Oncogene (2003) 22, 5181–5191& 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00

www.nature.com/onc

coordination or cognitive impairment and contrast-enhancing lesions may be observed in the white matterof the cortex upon magnetic resonance imaging. Theselesions represent areas of viral replication withinoligodendrocytes, the myelin-producing cells of theCNS. During the process of active viral infection, thehost oligodendrocytes succumb to lytic destruction.Treatment strategies for PML have aimed at controllingviral replication, but are currently at an experimentalstage and the prognosis remains poor for the majority ofaffected individuals.

Microscopically, the white matter lesions appear asareas of demyelination when sections of formalin-fixedparaffin-embedded tissue are stained with luxol fast blueand observed at low magnification (Figure 1a). Oligo-dendrocyte inclusion bodies, which contain enlargedglassy nuclei, are plentiful at the periphery of activelesions and represent cells in which viral replication isoccurring (Figure 1c). When the nuclear inclusionbodies are viewed by electron microscopy, dense sheetsof 45 nm icosahedral capsids consistent with polyoma-viridae size and structure can be seen (panel d).Interestingly, reactive or bizarre astrocytes are abun-dant, some of which may exhibit features observed intransformed cells such as multinucleation and pleo-morphism (Figure 1d).

Viral early protein, T-antigen

The polyomaviruses share a similar structure withgenomes composed of double-stranded circular DNAencoding the viral early proteins, T- and t-antigen, andthe viral late genes encoding the capsid proteins VP1,VP2, and VP3 as well as the accessory Agno protein.The early and late coding regions are separated by abidirectional regulatory region, which is the mostdivergent region among the three viruses. The viralregulatory protein, T-antigen, plays a critical role in theviral lifeycle in that it directs viral early and late geneexpression and viral DNA replication during lyticinfection (Major et al., 1992). In addition to its role inviral regulation during active replication, JCV T-antigenis considered an oncogene because of its demonstratedability to transform cells in culture. Although JCV doesnot appear to have the same efficiency of transformationas the well-described SV40 T-antigen, cells expressingJCV T-antigen exhibit characteristics of transformed orimmortalized cells including morphological changessuch as multinucleation, rapid doubling time, growthin anchorage independence, and subcutaneous growth inthe Nude mouse. JCV has successfully been used totransform rat fibroblasts and baby hamster kidney cells,albeit at low frequencies (Bollag et al., 1989; Haggerty

Figure 1 Histological features of PML. (a) Multiple areas of demyelination or plaques are observed at low magnification in paraffin-embedded sections of PML brain tissue stained with luxol fast blue. (b) Bizarre, transformed reactive astrocytes that may bemultinucleated and resemble neoplastic cells are frequently observed in PML lesions. (c) Residual JCV-infected oligodendrocytesharboring intranuclear eosinophilic inclusion bodies can be seen within demyelinated plaques. (d) Electron microscopy ofoligodendrocyte inclusions reveals the presence of 45 nm icosahedral viral particles in the nucleus consistent with JC virions(a: Luxol fast blue, original magnification � 100; b, c: hematoxylin and eosin, original magnifications � 400, � 1000, respectively;d: electron microscopy)

JCV and its role in carcinogenesisK Khalili et al

5182

Oncogene

et al., 1989). Indeed, JCV and SV40 T-antigens shareapproximately 70% sequence homology, with theC-terminal region that is responsible for host rangeand specificity being the most divergent. Similar toSV40, JCV T-antigen maintains helicase, a-polymerase,ATPase, and DNA-binding activities (Frisque et al.,1984). In addition, JCV T-antigen has been shown bycoimmunoprecipitation studies to interact physicallywith the tumor suppressor protein, p53 and theretinoblastoma protein family members, pRb, p130,and p107 (Bollag et al., 1989; Dyson et al., 1989, 1990;Ludlow and Skuse, 1995; Krynska et al., 2000). It isthrough binding that T-antigen is thought to sequesterand inactivate p53 and pRb, subsequently affectingnormal cell-cycle regulatory controls. While SV40 T-antigen has shown its ability to immortalize a broadrange of cells in vitro as well as to induce tumorigenesisin countless transgenic mouse lines, the transformingability of JCV T-antigen had appeared to be limited toneural-origin tissue based on a number of observationsin cell culture and animal models. However, more recentevidence has potentially expanded the range of cell typesto those outside the CNS, as outlined below, andsuggests more broad implications for the oncogenicpotential of JCV T-antigen.

Transmission, latency, and cellular tropism

JCV is considered a ubiquitous virus that is believed toinfect individuals early in life and establish a subacutechronic infection in the kidney. In support of thisnotion, earlier serological studies have demonstrated thedetection of circulating neutralizing antibody to JCV inapproximately 50% of the population under the age of20 reaching levels of greater than 70% of the adultpopulation (Padgett and Walker, 1973; Walker andPadgett, 1983). More recent reports have shown thedetection of JCV by PCR in the urine of nearly 50% ofhealthy individuals under the age of 30, while thepercentage of individuals shedding JCV into their urinesteadily increases with age (Kitamura et al., 1994).Owing to the prevalence of infection and the observa-

tion that JCV may be detected in tonsilar stromal cells(Monaco et al., 1998), viral transmission via therespiratory route has been hypothesized. However,recent evidence detecting viral DNA in the gastrointest-inal tract and the presence of viral genomic sequences inraw urban sewage suggest the possible fecal/oraltransmission of JCV and also emphasize the relativestability of viral particles (Ricciardiello et al., 2000;Bofill-Mas et al., 2001).

Insight regarding potential sites of latency in the bodymay come from infection experiments performed in cellcultures. Standard methods of culturing brain-derivedisolates of JCV have utilized primary human fetal glialcells, which are prepared from human fetal brain tissueand include populations of astrocytic, oligodendrocytic,as well as other cell types. Owing to the difficulty inobtaining human fetal tissue, and the inherent lack ofuniformity of primary cultures, initial cell culture studiesof JCV had been hampered by the lack of a stable cellline permissive for productive infection with JCV.However, several strains of JCV have been successfullypropagated in a number of primary cultures andestablished cell lines (Table 1). As might be anticipated,the virus is able to grow in the astrocytic cell lines, SVGand POJ, which were generated via transformation ofprimary fetal glial cells with replication-defective SV40and JC virus, respectively (Major et al., 1985; Mandlet al., 1987). Also, primary cultures of Schwann cells, themyelin-producing cells in the peripheral nervous system,are reported to be capable of supporting JCV replication(Assouline and Major, 1991). Of interest, in more recentstudies, several neuroblastoma cell lines have beenshown to support productive JCV infection (Akataniet al., 1994; Shinohara et al., 1997). The virus has alsobeen found to infect hematopoietic progenitor cells, Blymphocytes, and tonsilar stromal cells, although withmuch less efficiency (Monaco et al., 1998). With fewexceptions, the cell lines capable of supporting produc-tive JC viral infection are of human origin. The ability ofJCV to infect human cells may be restricted at the levelof viral DNA replication via the compatibility of T-antigen with the host synthesis machinery. In support ofthis notion, domain swapping of the carboxy termini of

Table 1 Cell culture systems that are susceptible to infection with JC virus

Cell origin Cell type Species Viral strain Propagation Reference

Glial AHB Human Productive Wroblewska et al. (1980)PHFG Human Mad-1, Mad-4 Productive Major and Vacante (1989)

Astrocytic POJa Human Productive Mandl et al. (1987)SVGb Human Mad-1 Productive Major et al. (1985) and Frye et al. (1997)

Neuroblastoma IMR-32 Human Mad-1, Tokyo-1 Productive Akatani et al. (1994)SH-EP Human JCV586 Productive Shinohara et al. (1997)SK-N-SH Human JCV586 Productive Shinohara et al. (1997)

Schwann PHFS Human Assouline and Major (1991)B-cell lymphoma BJAB Human Limited Atwood et al. (1992)

Namalwa Human Limited Atwood et al. (1992)Kidney epithelial COS-7b Monkey Archetype Productive Hara et al. (1998)

aTransformed with origin-defective JCV, expresses JCV T-antigen. bTransformed with origin-defective SV40, expresses SV40 T-antigenPHFG, primary human fetal glial; AHB, adult human brain; PHFS, primary human fetal Schwann cells.

JCV and its role in carcinogenesisK Khalili et al

5183

Oncogene

JCV and SV40 T-antigen, the region responsible forhost-range specificity, enables JCV DNA replication innon-human primate cells (Lynch et al., 1994). Of note,the Archetypal strain of JCV, which is derived from thekidney, can be propagated in the SV40-transformed cellline, COS-7, which was derived from African Greenmonkey kidney tissue. One may hypothesize that thepresence of high levels of SV40 T-antigen may enableJCV to overcome the species-specific blockage andpermit replication of the viral DNA. However, usingchimeric regulatory regions in which SV40 enhancersequences are inserted into the JCV promoter mayextend the host range of JCV to primate cells (Vacanteet al., 1989). Thus, it is likely that the host range andcell-type specificity of JCV may be determined by theviral regulatory region, in addition to T-antigen codingsequences.

Laboratory animal studies

Earlier efforts in animals focused on the development ofa suitable in vivo model of PML. Toward this end, JCVstrains isolated from the brains of individuals with PMLwere inoculated into newborn Syrian golden hamstersby various routes including intracerebral, intraocular,and intraperitoneal. Significant numbers of the animalsdeveloped neuronal and glial-origin tumors, mostfrequently medulloblastoma, peripheral neuroblastoma,astrocytoma, and primitive neuroectodermal tumors(Walker et al., 1973; Zu Rhein, 1983). Detailed analysesof less frequent tumors arising in hamsters highlight abroad collection of neoplasias including ependymoma,pineocytoma, pituitary carcinoma, retinoblastoma,choroid plexus papilloma, hemangiosarcoma, and ma-lignant schwannoma (Zu Rhein, 1983; Brun andJonsson, 1984; Nagashima et al., 1984). Similar studiesperformed using newborn Sprague–Dawley rats with theTokyo-1 strain of the virus resulted in the induction ofprimitive neuroectodermal tumors (Ohsumi et al., 1986).When examined by immunohistochemical analysis, thetumor tissues showed the expression of the viralT-antigen in the nucleus of the tumor cells. However,attempts to detect signs of viral replication including theexpression of the viral late capsid proteins by immuno-histochemistry or virion formation by electron micro-scopy were unsuccessful. No signs of myelin pathologyin any of the rodent models described above have beenreported in the literature.

Additional experiments were then designed to deter-mine the effect of age at the time of viral inoculation.Golden Syrian hamsters inoculated intracerebrally withvirus shortly after birth were found to develop tumors atmuch higher frequencies than animals inoculated at 2weeks of age (Matsuda et al., 1987). Furthermore,animals inoculated at birth had a much greater like-lihood for developing cerebellar medulloblastomas thananimals inoculated later in age, who more frequentlydeveloped glial-origin tumors in the cortex. Severalstudies have reported the appearance of microscopic

tumors in the internal granular layer of the cerebellumof inoculated newborn hamsters and have suggested thatcells of the external granular layer, which are activelyreplicating and migrating inward during early develop-ment, may be vulnerable to initial infection with JCV(Zu Rhein and Varakis, 1979; Nagashima et al., 1984;Matsuda et al., 1987). It is interesting to note that thecells of the external granular layer originate in the neuralcrest and are believed to give rise to medulloblastoma inhumans. In fact, the majority of the neoplasias describedin the hamster and rat models are tumors arising fromcells of the neural crest, which include medulloblastoma,primitive neuroectodermal tumors, glial-origin tumors,as well as others.

A number of experiments using New World primates,owl and squirrel monkeys, inoculated with purified JCVhave also been detailed in the literature. The majority ofthe monkeys were inoculated intracerebrally as adults,and a small subset received immunosuppressive drugsprior to and following inoculation to aid in tumorinducement (London et al., 1978). Several of the infectedanimals, with and without immunosuppression, devel-oped glial neoplasias in adulthood, including glioblas-toma multiforme and astrocytoma. With one exception,no evidence of viral replication was observed, althoughthe expression of the early gene, T-antigen, was detectedby immunohistochemistry or by Western blot analysis.Cells cultured from an astrocytoma arising in one owlmonkey inoculated with JCV were found to producespontaneously infectious JC viral particles, althoughsome rearrangement of the viral regulatory region mayhave occurred (Major et al., 1987). In all, JCV remainsthe only human virus reported to induce solid tumors innon-human primates.

Transgenic mouse models

Through the use of transgenic mice containing the genefor JCV T-antigen under the control of the viral earlypromoter, the role of T-antigen can be studied in theabsence of sequences encoding the capsid proteins. Anumber of independent lines of transgenic mice havebeen generated to date, which exhibit distinct pheno-types, as summarized in Table 2. Occasionally, founderanimals exhibiting signs of poor motor coordination andshaking consistent with the behavior seen in sponta-neous mutant models of demyelination have beenobserved in mice generated with the Mad-4 strain ofthe virus, suggesting alterations in myelin sheathformation. Indeed, electron microscopy revealed a lackof myelin sheath surrounding axons of the spinal cordand the presence of noncompacted myelin in twoindependent lines of mice (Small et al., 1986a, b).Subsequent molecular studies have suggested that JCVT-antigen may directly affect myelin gene expression atthe transcriptional level (Haas et al., 1994; Devireddyet al., 1996), while the possibility of an indirect effectfrom neighboring T-antigen expressing astrocytes hasbeen suggested from the analysis of mouse polyomavirus T-antigen transgenic mice with dysmyelination

JCV and its role in carcinogenesisK Khalili et al

5184

Oncogene

(Baron-Van Evercooren et al., 1992). The dysmyelinat-ing phenotype has been observed in several founderanimals, but its severity has prevented their successfulbreeding.

In addition to deficits in myelin sheath formation, thesame construct has been found to induce peripheralneuroblastomas and adrenal neuroblastomas on twodifferent genetic backgrounds. Using the Archetype, orkidney-derived isolate of JCV as the source of thetransgene, mice developed cerebellar tumors resemblinghuman medulloblastoma. Similar to peripheral neuro-blastomas, these very cellular tumors have a high mitoticindex, contain scant cytoplasm, and exhibit positivityfor the neuronal marker synaptophysin (Figure 2, panelsa and b). Approximately 80% of the animals developmedulloblastomas between 9 and 12 months of age. Thetumors impede motor coordination and balance usuallyresulting in hemiparesis as the overt phenotype.Immunohistochemical studies have demonstrated thatthe majority of tumor cells contain high levels of T-antigen in the nucleus (Figure 2, panel c). Westernblotting and immunoprecipitation/Western blot analysishave revealed high levels of T-antigen protein in thetumor tissue in association with the tumor suppressorproteins, p53 and pRb (Krynska et al., 1999a, b). Cellscultured from the tumor tissue have been separated intoclonal populations of T-antigen-positive and T-antigen-negative cells with only T-antigen-positive cell linesshowing the ability to grown subcutaneously in Nudemice (Krynska et al., 2000). Molecular analysis revealedthat the T-antigen-negative cell lines harbor a pointmutation affecting an intron–exon junction, whichresults in the expression of a smaller in-frame p53mRNA species lacking exon 4 (Krynska et al., 2000).Interestingly, this region of p53 is located in closeproximity to the T-antigen binding site, and thereforesuggests that tumor cells may require the presence ofT-antigen for inactivation of p53, while cells in whichmutant p53 variants are expressed may no longerrequire T-antigen expression.

In another series of transgenic mouse studies usingthe Mad-4 JCV isolate as the source of the T-antigentransgene in C57BL/6 mice, two different tumorphenotypes were observed to arise in the same lineof mice. The appearance of pituitary adenomas inthe transgenic mice appeared at high frequency with apenetrance of approximately 75% in mice greater than1 year of age (Gordon et al., 2000). Large hemorrhagicintracranial masses were frequently observed,which microscopically exhibited a high degree of

pleomorphism as well as a subset of prolactin-producingcells indicating the pituitary gland as the origin ofthe tumors (Figure 2, Panels d and e, respectively).These tumors had strong nuclear positivity forT-antigen in approximately 50% of the tumorcells (panel f). This same line of JCV T-antigentransgenic mice occasionally developed large, well-circumscribed masses in the scapular region oralong the extremities. Hematoxylin and eosin stainingrevealed interdigitating spindle cells and expressionof the neuronal marker protein, S-100 (Figure 2, Panelsg and h, respectively), consistent with malignantperipheral nerve sheath tumors (mpnst) (Gordon et al.,unpublished observations). Interestingly, the incidenceof pituitary tumors and mpnst mimics that observedin the human population in that pituitary tumors havea relatively high incidence as they account for 8–10% ofall intracranial neoplasms, while mpnst are fairlyuncommon (Leon et al., 1994; Woodruff et al., 2000).Observations similar to those made in the mousemedulloblastoma tumors, namely the expression ofT-antigen in the tumor tissue and its association withp53 and pRb, have been made in the pituitary tumorsand mpnst as well. Consistent in all the murine tumorswere a majority of cells exhibiting intense nuclearstaining for T-antigen and a subset of cells which appearnegative for T-antigen by immunohistochemistry(Figure 2, panels c, f, and i).

Association of JCV with human brain tumors

While previous case reports in the literature noted theassociation of CNS tumors in individuals with PML,most of these early cases appear circumstantial. Initialdescriptions of PML by EP Richardson reported thedisorder as a ‘heretofore unrecognized complication oflymphatic leukemia or Hodgkin’s disease’, and the whitematter lesions were generally discovered as incidentalfindings upon autopsy (Richardson, 1961). In themajority of the cases of concomitant PML and CNStumors, the patients developed an immunosuppressivedisease such as CNS lymphoma or were renal transplantrecipients receiving immunosuppressive drugs to controlgraft rejection. Two cases of glial tumors with con-comitant PML have been documented, one describingan individual with an ‘immunodeficiency syndrome’,and the other with no apparent underlying immunosup-pressive disease (Castaigne et al., 1974; Sima et al.,1983). However, in these cases, as well as the CNS

Table 2 Transgenic mouse models induced by expression of JCV T-antigen

Phenotype Promoter Mouse strain Reference

Dysmyelination Mad-4 FVB/N, CD-1 Small et al. (1986a, b) and Franks et al. (1996)Adrenal neuroblastoma Mad-4 CD-1 Small et al. (1986a, b)Peripheral neuroblastoma Mad-4 FVB/N Franks et al. (1996)Medulloblastoma Archetype FVB/N Krynska et al. (1999a, b)Pituitary adenoma Mad-4 C57BL/6J Gordon et al. (2000)Mpnst Mad-4 C57BL/6J Gordon (unpublished observations)

JCV and its role in carcinogenesisK Khalili et al

5185

Oncogene

lymphomas described above, viral replication was notedonly in the adjacent PML lesions, but not within thetumor itself.

In more recent years, mounting evidence has sug-gested that JCV may be associated with humanneoplasms in the absence of immunosuppression orPML. Through the use of sensitive molecular techniquesand the availability of more reliable immunochemicalreagents, JC viral genomic sequences and expression ofT-antigen have been observed in a variety of humantumor cell types. Rencic et al. (1996) reported the firstcase in which viral DNA and expression of JCVT-antigen were detected in the oligodendroglial portionof an oligoastrocytoma by PCR/Southern blot hybridi-zation, immunohistochemistry, and Western blotting.Subsequent analyses of oligodendrogliomas have re-vealed viral DNA in 20–75% and T-antigen expressionin up to 50% of tumors, as detailed in Table 3(Caldarelli-Stefano et al., 2000; Del Valle et al., 2001;Enam et al., 2002).

In addition to tumors of oligodendrocytes,the presence of JCV has been investigated in a broadrange of glial-origin tumors including astrocyto-mas, anaplastic astrocytic, anaplastic oligodendroglial,as well as glioblastomas and ependymomas. Asshown in Table 3, JCV DNA has been detectedin nearly every known type of malignant andnonmalignant tumor which arise from cells of theCNS. It is of interest that low-grade tumors suchas oligodendrogliomas are as likely to contain JCVDNA or express JCV T-antigen as a WHO Grade IVtumor such as a glioblastoma multiforme (Figure 3,Panels d, e, f). As such, it does not seem thatthe detection of JCV can be correlated with theaggressiveness or malignancy of the tumor, nor the typeof glial tumor. It should be noted that viral DNA maybe detected more frequently than JCV T-antigenexpression in the tumor tissues studied. Thus, the viralgenome, while it may be present, may not be expressedat levels detectable by immunohistochemical analysis.

Figure 2 Tumors observed in JCV T-antigen transgenic mouse models. Transgenic mice containing the complete sequences for JCVT-antigen under the control of the Archetype strain of the JCV early promoter develop medulloblastomas that are composed of smallhomogeneous sheaths of cells with scant cytoplasm (panel a) and express early neuronal markers such as synaptophysin (panel b). Micethat carry the JCV Mad-4-derived early promoter upstream of JCV T-antigen coding sequence develop adenomas arising from thepituitary gland (panel d) of which some cells are secretory and immunopositive for prolactin (panel e). The same JCVMad-4 T-antigentransgenic may also develop mpnst at a much lower frequency than pituitary adenomas. The mpnst are characterized by interdigitatingspindle cells (panel g) and express the neuronal marker, S-100 (panel h). All of the tumor types display strong nuclear staining in asubset of cells, as shown in the medulloblastoma (panel c), pituitary adenoma (panel f), and mpnst (panel i). (panels a, d, and g:hematoxylin and eosin staining, original magnification � 400; panels b, c, h, i: hematoxylin counterstain, original magnification � 400;panels e and f: hematoxylin counterstain, original magnification � 1000)

JCV and its role in carcinogenesisK Khalili et al

5186

Oncogene

Alternatively, the viral early coding region simply maynot be expressed.

Taking cues from the tropism of JCV in transgenicmice, a series of human medulloblastomas wereevaluated for the presence of JCV. Medulloblastomasare the most common malignant brain tumor in childrenwith nearly 70% appearing in children under the age of16 and rarely occurring in individuals over 50 (Molenaarand Trojanowski, 1994). While a small percentage arerelated to genetically defined heritable syndromesassociated with neoplasias, the majority are sporadicand of unknown etiology. A number of mutant geneshave been detected in medulloblastoma, including thehuman homologue of patched (PTCH) and b-catenin ofthe Wnt signaling pathway. Infrequently, mutations inthe p53 gene are found. In two separate studies of 43well-characterized medulloblastomas, nearly 77% wereshown to contain viral DNA corresponding to theN-terminal region of JCV T-antigen, while 36% of thetumors analysed by immunohistochemistry showedregions of the tumor with nuclear positivity forT-antigen (Figure 3, Panels a, b, c). In one study, theexpression of the JCV late protein, Agnoprotein, wasdetected in 69% of tumor samples (Del Valle et al.,2002a). Agno is an auxiliary protein whose functionremains unclear; however, recently it has been shown tobe expressed early in viral infection and can interactwith T-antigen to regulate viral transcription andreplication (Safak et al., 2001). Interestingly, in humantumor samples, Agnoprotein has been found to beexpressed in the cytoplasm of tumor cells and itsexpression does not appear to correlate with theexpression of T-antigen. Most recently, a role forAgnoprotein in dysregulation of the cell cycle by bindingto p53 and upregulating transcription of the p21/WAF-1promoter has been described (Darbinyan et al., 2002).However, further experiments are needed to determinewhat functional role, if any, that Agnoprotein may playin human tumorigenesis.

Association of JCV with non-neural human tumors

Owing to the tropism of the virus to non-CNSorigin tissues such as kidney, and the potential fornormal healthy individuals to excrete JCV in their urine,surveys of waste water were analysed for JCV andseveral studies have reported the detection of JCV viralparticles in raw sewage in a number of urban areas(Kitamura et al., 1994; Bofill-Mas et al., 2001). In light ofthe potential fecal-oral transmission of JCV, researchershave determined that JCV DNA is present in the upperand lower human gastrointestinal tract (Riccardielloet al., 2000, 2001). To follow-up on these observations,a series of adenocarcinomas of the colon were surveyedby PCR/Southern blot hybridization and immunohisto-chemistry for the presence of JCV. In one study, 83%of 29 tumor samples analysed were found to containviral DNA sequences (Laghi et al., 1999), while a secondstudy detected viral DNA in 81% of 27 samplesand nuclear T-antigen by immunohistochemistry in63% of the samples (Figure 3, Panels g,h,i)(Enam et al., 2002). In addition, approximately 44%of the colorectal tumors also contained evidence ofJCV Agnoprotein as well (Enam et al., 2002). It isinteresting to note that the colon tumors containJCV DNA and express the viral T-antigen protein atvery high percentages, at similar percentages as seen inthe CNS-origin tumors. The prevalence of JCVsequences and T-antigen in these tumors, along withthe human glial and medulloblastomas origin tumorsdescribed above suggests that these cell typesare permissive for the expression of the JCV earlypromoter, which may lead to the accumulation ofT-antigen and potential cellular transformation. Thisevidence, taken together with other potential sites oflatency in humans, including B cells and kidneyepithelium, suggest areas of further study to investigatethe presence of JCV in normal and tumor tissues of non-CNS origin.

Table 3 Detection of JC virus T-antigen DNA sequences and T-antigen expression in human tumors

Tumor type PCR/Southern IHC Strain Reference

Oligoastrocytoma 1/1 (100%) 1/1 (100%) Mad-4 Rencic et al. (1996)Xanthoastrocytoma 1/1 (100%) ND Mad-4 Boldorini et al. (1998)Medulloblastoma 20/23 (87%) 4/16 (25%) ND Krynska et al. (1999a,b)Colorectal cancera 24/29 (83%) ND Mad-1 Laghi et al. (1999)Astrocytoma 4/10 (40%) 1/10 (10%) Mad-4b Caldarelli-Stefano et al. (2000)Ependymoma 1/5 (20%) 0/5 (0%) Mad-4 Caldarelli-Stefano et al. (2000)Oligodendroglioma 1/5 (20%) 0/5 (0%) atypicalc Caldarelli-Stefano et al. (2000)Oligodendrogliomad 6/10 (60%) 4/14 (29%) ND Del Valle et al. (2001)Astrocytomab 18/21 (86%) 8/26 (31%) ND Del Valle et al. (2001)Oligoastrocytomad 7/11 (64%) 4/11 (36%) ND Del Valle et al. (2001)Glioblastoma 12/21 (57%) 5/26 (19%) ND Del Valle et al. (2001)Ependymomae 5/6 (83%) 5/6 (83%) ND Del Valle et al. (2001)Medulloblastoma 13/20 (65%) 9/20 (45%) ND Del Valle et al. (2002a)Glioblastoma 1/1 (100%) 1/1 (100%) Mad-4 Del Valle et al. (2002b)Oligodendroglioma 15/20 (75%) 10/20 (50%) Mad-4 Del Valle et al. (2002c)Colorectal carcinoma 22/27 (81%) 17/27 (63%) ND Enam et al. (2002)

aPCR amplification preceeded by topoisomerase I treatment. bRepresents the regulatory region of the IHC-positive sample. cRegulatoryrearrangement not related to any known strain. dIncludes low-grade and anaplastic tumors. eIncludes one subependymoma.ND¼not determined.

JCV and its role in carcinogenesisK Khalili et al

5187

Oncogene

Mechanisms of T-antigen-mediated transformation

A wealth of information concerning the potentialmechanisms by which T-antigen may lead to cellulartransformation has been obtained through in vitroanalysis as well as through the study of tissues fromT-antigen-induced tumors in experimental animal mod-els. As described above, the dysregulation of cell-cyclepathways through physical interaction with the tumorsuppressor proteins, p53 and pRb, is a common themein the majority of the animal models and has been welldescribed in cell culture systems. However, there arelikely additional levels at which T-antigen may interferewith cellular functions. Interestingly, in several of thetumor tissues, the expression of p21/WAF-1 was foundto be high in tumor cells where p53 is sequestered byT-antigen, indicating potential p53-independent activa-tion of p21/WAF-1 (Gordon et al., 2000).

Additional mechanisms by which T-antigen may exertcontrol over cell proliferation have been observedthrough studying signaling pathways in mouse medullo-blastoma cells. As a subset of human medulloblastomas

have shown deregulation of the Wnt signaling pathway,T-antigen-positive and -negative mouse medulloblasto-ma cell lines were evaluated for their levels of b-catenin,its cellular partner, LEF-1, and their activity on theirdownstream target, c-myc. Of interest, enhanced levelsof b-catenin and LEF-1 and a subsequent increase inc-myc levels were observed in T-antigen-positive cellscompared with T-antigen-negative cells (Gan et al.,2001). In another recent line of study, a role for theinsulin-like growth factor I receptor (IGF-IR) signalingsystem has been observed in the T-antigen-positivemedulloblastoma cell lines. More specifically, physicalinteraction of T-antigen with the insulin receptorsubstrate 1 (IRS-1), the major signaling molecule ofIGF-IR, and its nuclear localization in human medullo-blastoma biopsies suggest an additional signaling path-way that may be deregulated by JCV T-antigen (DelValle et al., 2002d).

In addition to its effect on the cell cycle, JCVT-antigen has been shown to possess mutagenic activityand may contribute to chromosomal instability ob-served in B lymphocytes (Theile and Grabowski, 1990;

Figure 3 Histological and immunohistochemical evaluation of human tumors harboring JCV. Human tumors characterized by lightmicroscopy include medulloblastoma (panel a), glioblastoma multiforme (panel d), and adenocarcinoma of the colon (panel g).Tumors were further characterized by immunohistochemistry for their expression of cell-type-specific markers confirming the tissue oforigin including the neuronal marker, synaptophysin, expressed by medulloblastomas (panel b), the glial-specific marker, GFAP (panele), and the epithelial marker cytokeratin for adenocarcinomas (panel h). Immunohistochemical staining for JCV T-antigen showspositive staining in the nuclei of tumor cells in all three tumor types (panels c, f, and i, respectively). (panels a, d, and G: hematoxylinand eosin staining; panels b, c, e, f, h, and i: hematoxylin counterstain; All panels: original magnification � 400)

JCV and its role in carcinogenesisK Khalili et al

5188

Oncogene

Lazutka et al., 1996; Neel et al., 1996). In furtherevidence of the ability of T-antigen to induce alterationsat the chromosome level, morphological changes such asmultinucleation, high nuclear pleomorphism, aneuploi-dy, and polyploidy have been observed in T-antigen-induced transgenic mouse tumors and cell culturesderived from those tumors (Krynska et al., 2000;Gordon et al., unpublished observations). Once dysre-gulation of one or more cellular pathways or inductionof chromosomal instability occurs, shifts in cell-cycleprofiles or genetic mutations may render T-antigenexpression unnecessary. Such a ‘hit-and-run’ or amultistep transformation strategy has been suggestedfor cancers such as colorectal carcinoma. The observedloss of T-antigen expression in subpopulations ofneural-origin tumor cells suggests that T-antigen mayfunction at an earlier rather than later step intumorigenesis.

Discussion

It should be noted that transcriptional activation of theJCV early promoter in a given cell type is necessary forT-antigen expression. When JCV early gene expression

is followed by viral DNA replication and late geneexpression, a productive infection results. As is the casewith PML, JCV-infected oligodendrocytes are permis-sive for viral lytic infection and are destroyed in theprocess, leading to demyelination. However, one canenvision a scenario in which an infected cell may bepermissive for early gene expression, although not ableto support viral replication or late gene expression. Inthis setting, the expression of T-antigen may lead toinactivation of tumor suppressors, dysregulation ofsignaling pathways, or DNA instability, which couldcontribute to transformation and subsequent tumori-genesis. Factors contributing to the permissiveness of agiven cell type may lie in its tissue of origin, point ofdevelopment, or state of differentiation. For example,perhaps it is the mature oligodendrocyte that can fullysupport a viral lytic infection while an undifferentiatedoligodendrocyte may only allow T-antigen expression,leading to the development of an oligodendroglioma.Given the well-established tissue tropism of JCV towardneural-origin cells and the predilection in animal modelsfor induction of tumors of neural crest origin, one mayanticipate the detection of JCV in neuronal and glial-origin tumors. However, more recent evidence demon-strating the presence of the virus in sewage and in

JCVProductiveinfection

Abortiveinfection

Permissive cells(oligodendrocytes)

Expression of viral T-antigenReplication of viral DNA

Expression of viral capsidViral assembly

Cytolytic destruction of infected cells

PML

AstrocytesNeuronal Cells

Intestinal epitheliumOther tissues

Expression of viral T-antigenNo viral DNA replication

No virion formation

T-antigen

Dysregulation ofsignaling pathways

i.e. Wnt, IGF

Inactivation oftumor suppressors,

i.e. p53, pRb

Interfering withDNA repair

Uncontrolled cell proliferationtumorigenesis

Tumor

Figure 4 Mechanisms of JCV T-antigen-mediated cellular transformation or demyelination

JCV and its role in carcinogenesisK Khalili et al

5189

Oncogene

normal and cancerous human colon tissue may require are-evaluation of the tropism of JC virus.

The growing body of literature detecting JCV, as wellas SV40 and BKV sequences, in normal and pathologichuman tissues raises several important issues, whichmust be carefully considered. Firstly, the presence ofviral genomic DNA in itself does not establish causa-tion. It is necessary to show the expression of viral genesin order suggest their participation in aberrant cellularproliferation. However, as T-antigen expression may notbe required for maintenance of a transformed state, itmay be difficult to establish whether or not polyomavir-al DNA present in a tumor did in fact result in theproduction of viral proteins. In any case, T-antigen,when expressed, has well-established oncogenic proper-ties and is capable of targeting a number of key cellularregulatory pathways and it is likely that additional

cellular pathways affected by T-antigen expression willbe uncovered by future in vitro and in vivo studies. Assuch, mechanisms of T-antigen-mediated transforma-tion can be studied in vitro and through the use oflaboratory animal models to provide better under-standing of cellular regulatory pathways involved intumorigenesis. Such studies may lead to the identifica-tion of therapeutic targets in JCV-transformed cells,potentially even the targeting of T-antigen itself.

Acknowledgements

We thank past and present members of the Center forNeurovirology and Cancer Biology for their insightful discus-sion, and sharing of ideas of reagents. We also thank CSchriver for preparation of this manuscript. This work wasmade possible by grants awarded by NIH to JG and KK.

References

Akatani K, Imai M, Kimura M, Nagashima K and Ikegami N.(1994). J. Med. Virol., 43, 13–19.

Arrington AS and Butel JS. (2001). Human Polyomaviruses:Molecular and Clinical Perspectives, Khalili K, Stoner GL(eds). Wiley-Liss: New York, NY.

Assouline JG and Major EO. (1991). J. Virol., 65, 1002–1006.Atwood WJ, Amemiya K, Traub R, Harms J and Major EO.(1992). Virology, 190, 716–723.

Baron-Van Evercooren A, Jensen NA, Wyss MT, Cuzin F,Rassoulzadegan M, Brucher JM and Baron H. (1992). Lab.Invest., 66, 39–53.

Berger JR and Concha M. (1995). J. Neurovirol., 1, 5–18.Bofill-Mas S, Formiga-Cruz M, Clemente-Casares P,Calafell F and Girones R. (2001). J. Virol., 75, 10290–10299.

Boldorini R, Caldarelli-Stefano R, Monga G, Zocchi M,Mediati M, Tosoni A and Ferrante P. (1998). J. Neurovirol.,2, 242–245.

Bollag B, Chuke WF and Frisque RJ. (1989). J. Virol., 63,

863–872.Brun A and Jonsson N. (1984). Cancer, 53, 1714–1717.Caldarelli-Stefano R, Boldorini R, Monga G, Meraviglia E,Zorini EO and Ferrante P. (2000). Hum. Pathol., 3,

394–395.Castaigne P, Rondot P, Escourolle R, Ribadeau Dumas J-L,Cathala F and Hauw J-J. (1974). Rev. Neurol. (Paris), 130,

379–392.Corallini A, Tognon M, Negrini M and Barbanti-Brodano G.(2001). Human Polyomaviruses: Molecular and ClinicalPerspectives, Khalili K, Stoner GJ (eds).. Wiley-Liss: NewYork, NY.

Darbinyan A, Darbinian N, Safak M, Radhakrishnan S,Giordano A and Khalili K. (2002). Oncogene, 21,

5574–5581.Del Valle L, Delbue S, Gordon J, Enam S, Croul S, Ferrante Pand Khalili K. (2002b). Neurology, 58, 895–900.

Del Valle L, Enam S, Lara C, Ortiz-Hidalgo C, KatsetosCD and Khalili K. (2002c). Clin. Cancer Res., 8,

3332–3340.Del Valle L, Gordon J, Assimakopoulou M, Enam S, GeddesJF, Varakis J, Katsetos CD, Croul S and Khalili K. (2001).Cancer Res., 61, 4287–4293.

Del Valle L, Gordon J, Enam S, Delbue S, Croul S,Abraham S, Radhakrishnan S, Assimakoupoulou M,Katsetos CD and Khalili K. (2002a). J. Natl. Cancer Inst.,94, 267–273.

Del Valle L, Wang JY, Lassak A, Peruzzi F, Croul S, Khalili Kand Reiss K. (2002d). J. Neurovirol., (Suppl. 2) 8, 138–147

De Mattei M, Martini F, Corallini A, Gerosa M, Scotlandi K,Carinci P, Barbanti-Brodano G and Tognon M. (1995). Int.J. Cancer, 61, 756–760.

Devireddy LR, Kuman KU, Pater MM and Pater A. (1996).J. Med. Virol., 49, 205–211.

Dyson N, Bernards R, Friend SH, Gooding LR, Hassel JA,Major EO, Pipas JM, Van Dyke T and Harlow E. (1990).J. Virol., 64, 1353–1356.

Dyson N, Buchkovich K, Whyte P and Harlow E. (1989). Cell,58, 249–255.

Enam S, Del Valle L, Lara C, Gan DD, Ortiz-Hidalgo C,Palazzo JP and Khalili K. (2002). Cancer Res., 62, 7093–7101.

Franks RR, Rencic A, Gordon J, Zoltick PW, Curtis M,Knobler RL and Khalili K. (1996). Oncogene, 12, 2573–2578.

Frisque RJ, Bream GL and Cannella MT. (1984). J. Virol., 51,

458–469.Frye S, Trebst C, Dittmer U, Petry H, Bodemer M, HunsmannG, Weber T and Luke W. (1997). J. Virol. Methods, 63,

81–92.Gan DD, Reiss K, Gorrill T, Del Valle L, Croul S, GiordanoA, Fishman P and Khalili K. (2001). Oncogene, 20,4864–4870.

Gordon J, Del Valle L, Otte J and Khalili K. (2000). Oncogene,19, 4840–4846.

Haas S, Haque NS, Beggs AH, Khalili K, Knobler RL andSmall J. (1994). Virology, 202, 89–96.

Haggerty S, Walker DL and Frisque RJ. (1989). J. Virol., 63,

2180–2190.Hara K, Sugimoto C, Kitamura T, Aoki N, Taguichi F andYogo Y. (1998). J. Virol., 72, 5335–5342.

Kitamura T, Kunitkake T, Guo J, Tominaga T,Kawabe K and Yogo Y. (1994). J. Clin. Microbiol., 32,2359–2363.

Krynska B, Del Valle L, Gordon J, Otte J, Croul S and KhaliliK. (2000). Virology, 274, 65–74.

JCV and its role in carcinogenesisK Khalili et al

5190

Oncogene

Krynska B, Del Valle L, Croul S, Gordon J, Katsetos K,Carbone M, Giordano A and Khalili K. (1999a). Proc. Natl.Acad. Sci. USA, 96, 11 519–11 524.

Krynska B, Otte J, Franks R, Khalili K and Croul S. (1999b).Oncogene, 18, 39–46.

Laghi L, Randolph AE, Chauhan DP, Marra G, Major EO,Neel JV and Boland CR. (1999). Proc. Natl. Acad. Sci. USA,96, 7484–7489.

Lazutka JR, Neel JV, Major EO, Dedonyte V, Mierauskine J,Slapsyte G and Kesminieni A. (1996). Cancer Lett., 109,177–183.

Leon SP, Zhu L and Black P McL. (1994). Neurosurgery, 34,

708–722.London WT, Houff SA, Madden DL, Fuccillo DA, GravellM, Wallen WC, Palmer AE, Sever JL, Padgett BL, WalkerDL, ZuRhein GM and Ohashi T. (1978). Science, 201, 1246–1249.

Ludlow JW and Skuse GR. (1995). Virus Res., 35, 113–121.Lynch KJ, Haggerty S and Frisque RJ. (1994). Virology, 204,

819–822.Major EO, Amemiya K, Tornatore CS, Houff SA and BergerJR. (1992). Clin. Microbiol. Rev., 5, 49–73.

Major EO, Miller AI, Mourrain P, Traub RG, De Widt E andSever J. (1985). Proc. Natl. Acad. Sci. USA, 82, 1257–1261.

Major EO and Vacante DA. (1989). J. Neuropathol. Exp.Neurol., 48, 425–436.

Major EO, Vacante DA, Traub RG, London WT and SeverJL. (1987). J Virol., 61, 1435–1441.

Mandl C, Walker DL and Frisque RJ. (1987). J. Virol., 61,

755–763.Matsuda M, Yasui K, Nagashima K and Mori W. (1987). J.

Natl. Cancer Inst., 79, 585–591.Molenaar WM and Trojanowski JQ. (1994). Crit. Rev. Oncol.

Hematol., 17, 1–25.Monaco MCG, Jensen PN, Hou J, Durham LC and MajorEO. (1998). J. Virol., 9918–9923.

Nagashima K, Yasui K, Kimura J, Washizu M, Yamaguchi Kand Mori W. (1984). Am. J. Pathol., 116, 455–463.

Neel JV, Major EO, Awa AA, Glover T, Burgess A, Troub R,Curfman B and Satoh C. (1996). Proc. Natl. Acad. Sci. USA,93, 2690–2695.

Ohsumi S, Motoi M and Ogawa K. (1986). Acta. Pathol. Jpn.,36, 815–825.

Padgett BL and Walker DL. (1973). J. Infect. Dis., 127, 467–470.Price RW. (1996). Lancet, 348, 445–452.Rencic A, Gordon J, Otte J, Curtis M, Kovatich A, Zoltick P,Khalili K and Andrews D. (1996). Proc. Natl. Acad. Sci.USA, 93, 7352–7357.

Riccardiello L, Chang DK, Laghi L, Goel A, Chang CL andBoland R. (2001). J. Virol., 75, 1996–2001.

Ricciardiello L, Laghi L, Ramamirtham P, Chang CL, ChangDK, Randolph AE and Boland CR. (2000). Gastroenterol-ogy, 119, 1228–1235.

Richardson EP. (1961). New Engl. J. Med., 265, 815–823.Safak M, Barrucco R, Darbinyan A, Okada Y, Nagashima Kand Khalili K. (2001). J Virol, 75, 1476–1486.

Shinohara T, Nagashima K and Major EO. (1997). Virology,2, 269–277.

Sima AAF, Finkelstein SD and MacLachlan DR. (1983). Ann.Neurol., 14, 183–188.

Small JA, Scangos GA, Cork L, Jay G and Khoury G.(1986a). Cell, 46, 13–18.

Small JA, Khoury G, Jay G, Howley PM and ScangosGA. (1986b). Proc. Natl. Acad. Sci. USA, 83, 8288–8292.

Theile M and Grabowski G. (1990). Arch. Virol., 113, 221–233.Vacante DA, Traub R and Major EO. (1989). Virology, 170,

353–361.Walker DL and Padgett BL. (1983). Prog. Clin. Biol. Res., 105,99–106.

Walker DL, Padgett BL, Zu Rhein GM, Albert AE and MarshRF. (1973). Science, 181, 674–676.

Woodruff JM, Kourea HP, Louis DN and Scheithauer BW.(2000). Pathology and Genetics of Tumors of the NervousSystem, Kleihues P, Cavenee WK (eds.). IARC Press: Lyon,France.

Wroblewska Z, Wellish M and Gilden D. (1980). Arch. Virol.,65, 141–148.

Zu Rhein GM. (1983). Polyomaviruses and Human NeurologicalDiseases. Alan R. Liss, Inc.: New York, NY, pp. 205–221.

Zu Rhein GM and Varakis JN. (1979). Natl. Cancer Inst.Monogr., 51, 205–208.

JCV and its role in carcinogenesisK Khalili et al

5191

Oncogene