|Home | About | Journals | Submit | Contact Us | Français|
Summary: Progressive multifocal leukoencephalopathy (PML) is a debilitating and frequently fatal central nervous system (CNS) demyelinating disease caused by JC virus (JCV), for which there is currently no effective treatment. Lytic infection of oligodendrocytes in the brain leads to their eventual destruction and progressive demyelination, resulting in multiple foci of lesions in the white matter of the brain. Before the mid-1980s, PML was a relatively rare disease, reported to occur primarily in those with underlying neoplastic conditions affecting immune function and, more rarely, in allograft recipients receiving immunosuppressive drugs. However, with the onset of the AIDS pandemic, the incidence of PML has increased dramatically. Approximately 3 to 5% of HIV-infected individuals will develop PML, which is classified as an AIDS-defining illness. In addition, the recent advent of humanized monoclonal antibody therapy for the treatment of autoimmune inflammatory diseases such as multiple sclerosis (MS) and Crohn's disease has also led to an increased risk of PML as a side effect of immunotherapy. Thus, the study of JCV and the elucidation of the underlying causes of PML are important and active areas of research that may lead to new insights into immune function and host antiviral defense, as well as to potential new therapies.
Before the AIDS pandemic and the use of immunomodulatory therapy, progressive multifocal leukoencephalopathy (PML) was an extremely rare disease, associated primarily with underlying neoplastic conditions causing a defect in immune function (20, 419). Interestingly, PML was associated mainly with B cell lymphoproliferative disorders (57, 198), which have been hypothesized to lead to the spread of virus from potential sites of latency to the brain. Accounts of potential cases of PML can be traced back as far as 1930 (29, 85, 181, 419, 537). The first case of demyelinating disease described with the term PML was found in a patient with chronic lymphocytic leukemia (CLL) and Hodgkin's lymphoma in 1958 (20). These cases are all consistent with the pathology of PML, including the development of multiple white matter plaques in the brain stem, basal ganglia and thalamus, cerebral hemispheres, and cerebellum.
A viral cause of PML was proposed in 1959 due to observations of inclusion bodies in the nuclei of damaged oligodendrocytes (70) and the hypothesis that the distribution of lesions could be explained by an atypical viral infection (419). The nuclei of cells with inclusion bodies were found by electron microscopy to contain particles similar to known polyomaviruses (202, 559, 560). The etiological agent of PML was not isolated until 1971, when the virus was isolated from a mixed culture of glial cells following a “blind” passage (380) and named JC virus (JCV), after the initials of the patient. More recently, JCV has been termed JC polyomavirus (JCPyV), but this review will maintain the more common nomenclature of “JCV.”
JCV was found to be a nonenveloped icosahedron of 40 nm diameter, which, unlike simian virus 40 (SV40), could cause hemagglutination (HA) of human type 0 erythrocytes (377), which provided means to perform seroepidemiological studies. Data from these studies indicated that JCV was found globally (58), that seroconversion of a large percentage of the population occurred before adulthood (431, 522), and that healthy people, including pregnant women, produced immunoglobulin G (IgG) against JCV (17, 99). Therefore, PML was likely to be caused by reactivation of a latent infection (57, 379, 522). For a full review of the historical association of JCV and PML, see reference 301 and references therein.
PML ceased to be a rare disease after HIV became widespread in the human population. Estimates of the occurrence of PML in AIDS patients range from 3 to 5% (299). The incidence of PML in AIDS patients is significantly greater than that in persons with other underlying causes of immunosuppression (299). Notably, PML incidence has decreased less significantly than other opportunistic infections since the advent of highly active antiretroviral therapy (HAART) (103, 546). It is unclear why PML occurs more frequently in AIDS patients than in those with other underlying causes of immunosuppression, although some causes may include changes in immune cell trafficking, blood-brain barrier (BBB) permeability and cytokine secretion, interaction between viral proteins in coinfected cells, and damage to the brain caused by neuronal HIV infection. HIV likely affects both the immune system and the local cellular environment in ways that increase the risk of progression to PML.
The development of PML as a side effect of immunomodulatory therapy is a growing concern, with reports of fatal PML cases in patients treated with natalizumab (Tysabri) for multiple sclerosis (MS) and Crohn's disease, with rituximab (Rituxan) for multiple sclerosis, non-Hodgkin's lymphoma, rheumatoid arthritis, autoimmune hematological disorders, myasthenia gravis, systemic lupus erythematosus (SLE), and B cell lymphoma, with efalizumab (Raptiva) for plaque psoriasis, with infliximab (Remicade) for psoriasis, Crohn's disease, ankylosing spondylitis, psoriatic arthritis, rheumatoid arthritis, and ulcerative colitis, and with mycophenolate mofetil (Cellcept) for suppression of organ transplant rejection (315). These therapies target immune cells, inhibiting their biological function. The risk of PML during natalizumab treatment rises as treatment progresses. The true overall incidence of PML due to natalizumab therapy is currently estimated at 3.85 per 1,000 patients (available for prescribing physicians at https://medinfo.biogenidec.com) and approximately 1 per 500 efalizumab patients. Prescription labeling for mycophenolate mofetil contains a warning of PML risk. Natalizumab and rituximab contain FDA-mandated “black box warnings” because of the risk of PML (http://www.accessdata.fda.gov/scripts/cder/drugsatfda/), while efalizumab has been voluntarily withdrawn from the market because of concerns about the frequency of occurrence of PML (299). Although these therapies are effective for their intended use, the incidence of PML, a deadly disease with no approved treatment, limits their use. Thus, continued research and a more thorough understanding of JCV biology, epidemiology, and pathology are of continued and increasing importance.
JCV, like all polyomaviruses, is a nonenveloped, T = 7 icosahedral virus with a closed circular, supercoiled, double-stranded DNA genome. The capsid is composed of three viral structural proteins, VP1, VP2, and VP3, with VP1 being the major constituent. There are 72 pentamers, each composed of five VP1 molecules and one molecule of either VP2 or VP3 (302). Only VP1 is exposed on the surface of the capsid, and it thus determines receptor specificity. Polyomavirus DNA is nucleosomal in structure, with approximately 25 nucleosomes composed of viral DNA and host cell histones contained in each minichromosome, as determined for SV40 (12, 264, 333). The viral particle does not contain linker histones, but the genome acquires them after entry into the host cell.
The prototype JCV genome (Fig. 1) is 5,130 bp (154), although individual variants differ in length, due to alterations in their noncoding regions (see Fig. 3). The genome encodes 6 major viral proteins (large T and small t antigens, VP1, VP2, VP3, and agnoprotein) as well as several splice variants of T antigen. Early- and late-transcribing sides of the genome are physically separated by a noncoding control region (NCCR), often called the hypervariable regulatory region (HVRR) or regulatory region (RR), and are transcribed in opposite directions from opposite strands of DNA. The early side of the genome, which is transcribed before DNA replication begins, is composed of large T antigen and small t antigen genes, as well as the splice variants T′135, T′136, and T′165. The late side of the genome is transcribed concomitant with DNA replication and encodes the three viral structural proteins, VP1, VP2, and VP3, as well as the accessory agnoprotein.
Large T antigen and its variants are multifunctional, interacting with both host and viral proteins as well as with both host and viral DNAs. The T proteins are involved in driving the host cell toward S phase for viral replication, regulating transcription from the host and viral genomes, and directly participating in viral DNA replication. The agnoprotein is less well studied but also appears to be multifunctional, and highly varied functions have been attributed to it (reviewed in reference 238), ranging from viral transcriptional regulation to inhibition of host DNA repair to functioning as a viroporin (478).
Polyomaviruses all display restricted species and cell type specificities for lytic infection. The strict species specificity of JCV for humans has confounded the development of animal models for PML. In vivo, JCV infection is likely restricted to kidney epithelial cells, tonsillar stromal cells, bone marrow-derived cell lineages, oligodendrocytes, and astrocytes (22, 201, 300, 301, 341, 342, 494). The virus is thought to establish low-level persistent or latent infections in the kidney and in bone marrow-derived cells largely due to inefficient viral replication in these cell types. Once in the central nervous system (CNS), the virus replicates vigorously in oligodendrocytes, leading to the demyelinating disease PML. The cell type-specific tropism of JCV observed in vivo is mirrored in vitro, with virus productively infecting bone marrow-derived cells, tonsillar stromal cells, and macroglia (142, 303, 306, 307, 344). Virus replication is maximal in primary human fetal glial cell cultures and in some human glial cell lines (303), as well as in other cell lines expressing SV40 large T antigen, such as COS-7 cells (185, 303). However, it has been our experience that JCV does not multiply well in COS-7 cells compared with the human SVG cell line, which was established with the same SV40 vector as the COS cell lines.
The major tropism of JCV for human glial cells is not fully understood, but multiple factors are likely responsible for contributing to robust viral replication in this cell type. Host cell- and species-specific transcription and replication factors contribute significantly to the restricted specificity displayed by JCV and other members of the genus (142, 292, 480). In addition, virus-receptor interactions contribute to viral tropism and spread. This was first shown for JCV using a JCV/SV40 hybrid virus that contained the NCCR and T antigen-coding genes of SV40 as well as the JCV capsid-coding genes (76). The hybrid virus induced an SV40-like cytopathic effect in human glial cells and hemagglutinated human type O red blood cells similarly to JCV (76).
In all polyomaviruses studied to date, initial binding to host cells involves an interaction with negatively charged sialic acid-containing receptors. The early binding and entry events in the related polyomaviruses SV40 and murine polyomavirus (mPyV) have been extensively studied and contribute to our understanding of JCV pathogenicity. The monosialylated ganglioside GM1 is the major receptor for SV40 (26, 63, 499). SV40 has also been shown to use the class I major histocompatibility complex protein 1 as a receptor in some cases (54, 366, 468). Initially, it was thought that SV40 was not able to attach to sialic acids, as neuraminidase treatment of host cells did not reduce infectivity, although it was later shown that neuraminidase fails to cleave sialic acid off GM1 (88, 337). This interaction between SV40 and GM1 is extremely specific, as SV40 preferentially binds to N-glycolyl GM1 over N-acetyl GM1, which differ by a single hydroxyl group (63). Additionally, although a single VP1 molecule binds GM1 weakly, the multivalent viral capsid is expected to bind GM1 molecules on the plasma membrane with an affinity that is several orders of magnitude greater (362).
Early studies with mPyV demonstrated that the virus binds to α2,3- and α2,6-linked sialic acids on host cells (61, 62, 149, 383). Subsequently, the mPyV was shown to bind to sialic acids present on the ganglioside receptors GD1a and GT1b (461, 499). Although mPyV utilizes gangliosides for productive infection, the virus will bind to sialic acids present on glycoproteins (397). Binding of mPyV to cell surface glycoproteins targeted the virus to a nonproductive pathway for degradation, suggesting that these glycoproteins represent nonproductive pseudoreceptors. The specific interactions between mPyV and sialic acids also play a critical role in pathogenicity. Early studies demonstrated that while both small- and large-plaque variants of mPyV were able to bind to α2,3-linked sialic acids, small-plaque variants were additionally able to bind α2,6-linked sialic acids due to a single point mutation in VP1 (62, 148, 469, 470). As a result, large-plaque variants bound to cells less tightly. This reduced binding also increased pathogenesis of mPyV in mice, indicating that the reduced binding to sialic acids increased viral spread (123, 148).
JCV also requires sialic acid to infect cells and has been reported to utilize both α2,3- and α2,6-linked sialic acids to infect permissive glial cells (125, 284). After stripping glial cells of α2,3- and α2,6-linked sialic acids, each linkage was reconstituted using linkage-specific sialyltransferases. Restoring either α2,3- or α2,6-linked sialic acids to cells rescued infection, indicating that both α2,3- and α2,6-linked sialic acids play an important role in infection (125). The sialyltransferases used were also specific for adding carbohydrate to N-linked glycoproteins, and as infection was fully restored, it was suggested that an N-linked glycoprotein containing either α2,3- or α2,6-linked sialic acid was sufficient for virus infection.
Recently, a receptor moiety used by JCV for productive infection has been identified. Using glycan arrays, a recombinant JCV VP1 pentamer has been shown to bind specifically to lactoseries tetrasaccharide C (LSTc), which contains a terminal α2,6-linked sialic acid (361). This unusual molecule adopts an “L” shape, and JCV specifically binds this molecule by interacting with not only the terminal α2,6-linked sialic acid but also the adjacent GlcNAc sugar. Comparing the crystal structures of SV40 in complex with a portion of GM1 and of JCV in complex with LSTc reveals that the α2,3 versus α2,6 linkages and the favorable interactions between JCV and GlcNAc largely determine binding specificity. Sialylparagloboside, which is identical to LSTc except for its terminal α2,3-linked sialic acid, does not bind JCV. Although α2,3-linked sialic acids were present on the glycan array, none demonstrated appreciable binding to the VP1 pentamer. When the multivalent JCV virus-like particles (VLPs) were used, low binding to some α2,3-sialic acids, including the gangliosides GM2 and GM1, was seen. Therefore, it appears likely that JCV primarily uses α2,6-linked sialic acids to bind to cells, with the possibility that the weak interactions between JCV and α2,3-linked sialic acids play a limited role in infection. It is currently unknown onto what receptor molecule the LSTc moiety is attached.
Viral DNA recovered from cerebrospinal fluid specimens (CSF) from patients with PML has been shown to code for amino acid substitutions of several VP1 residues responsible for binding to sialic acids (477). These mutations are predicted to reduce binding to sialic acids, although it is unclear whether these mutations will confer a selective advantage in patients, as similar mutations clearly reduce infectivity of JCV in tissue culture (362). Alternatively, these mutations may result in a selective advantage similar to that seen in mPyV by preventing nonproductive adsorption to cells that are not permissive for JCV infection. However, these mutations appear to block binding to both α2,3- and α2,6-linked sialic acids, and JCV VLPs containing these mutations show a reduced hemagglutination ability and a decreased ability to bind gangliosides (172, 361). Interestingly, these mutations were recovered only in the CSF and bloodstream and not in the same patient's urine. As only a single genotype of virus was recovered in each patient, it appears that these mutations arose from positive selection of a single JCV population (413). It is currently unclear whether these mutations arise in the CNS of PML patients or originate in peripheral sites before trafficking to the brain. Further studies will be necessary to determine the effects of these mutations on JCV pathogenesis.
In addition to using sialic acid as a receptor, JCV has been shown to require the serotonin receptor, 5HT2AR, to infect glial cells (130). The initial observations were centered around the fact that antipsychotic drugs such as chlorpromazine (Thorazine) and clozapine potently inhibit virus infection (30). As these drugs antagonize both serotonin and dopamine receptors, it was hypothesized that one or both of these neurotransmitter receptors might function as a virus receptor on glial cells. Subsequent studies using drugs that specifically antagonized dopamine or serotonin receptors suggested that the 5HT2A subtype of the serotonin receptor functioned as a JCV receptor (130, 373). It was then shown that expression of the 5HT2A receptor was sufficient to allow viral entry to cells lacking this receptor (HeLa and HEK293A) and that antibodies to the 5HT2AR blocked infection (130, 296). These data when taken together strongly support a role for 5HT2AR in JCV infection of glial cells. In vivo data demonstrate increased expression of the 5HT2A receptor in and around PML lesions, further supporting a critical role for this receptor in the pathogenesis of PML (unpublished observations). Interestingly, infection of 5HT2AR-expressing cells remains neuraminidase sensitive, confirming a prominent role for cellular carbohydrates in infection (unpublished observations). The 5HT2AR protein contains several potential glycosylation sites, but it is unclear whether LSTc or other sialic acid moieties responsible for infection reside on this protein. Elimination of glycosylation sites on 5HT2AR also prevents receptor expression of the cell surface, preventing a clear consensus on the need for sialic acids on 5HT2A for infection (296). Alternatively, LSTc may instead be present on other sialic acid-containing molecules. One group has found that JCV can infect human brain microvascular endothelial cells that lack the 5HT2A receptor (75). Infection in their system was very inefficient, but the results indicate that virus infection can proceed by alternative mechanisms on some cell types. Additionally, gangliosides may play a role in JCV infection, and the ganglioside GT1b has been reported to function as a receptor for JCV (248).
Unlike mPyV, SV40, or BK virus (BKV), JCV enters cells by clathrin-dependent endocytosis (389; reviewed in reference 500) (Fig. 2). Pharmacological inhibitors and expression of dominant negative proteins directed at clathrin-dependent endocytosis inhibit virus infection, whereas inhibition of caveola-dependent endocytosis has no effect on replication (389). Consistent with clathrin-mediated endocytosis, JCV traffics from clathrin-coated pits to Rab5-positive early endosomes, a step that is blocked by dominant negative Eps15 mutants (398) (Fig. 2). Additionally, JCV colocalizes with green fluorescent protein (GFP)-labeled Rab5, and expression of dominant negative inhibitors to Rab5 prevent infection. Subsequently, JCV colocalizes with cholera toxin B, a marker for lipid-mediated endocytosis used by SV40 and mPyV, in compartments that are most likely caveolin-1-positive late endosomes (traditionally referred to as caveosomes) (134, 140, 399). Surprisingly, JCV does not colocalize with markers for Rab7 late endosomes, and expression of the dominant negative form of Rab7 does not inhibit infection. This suggests that trafficking to Rab7-positive late endosomes is not necessary for infection and that virions may leave the normal endocytic pathway from Rab5-positive endosomes or Rab5- and Rab7-positive maturing endosomes. At 12 to 16 h postinfection, JCV colocalizes with the endoplasmic reticulum (ER) protein calregulin, suggesting that JCV traffics to the ER for productive infection. Treatment of cells with brefeldin A, which inhibits COP1-mediated ER trafficking, potently inhibits infection, which further suggests that trafficking of JCV to the ER is a critical step in infection (399).
There is probably much to be learned from SV40 and mPyV that has not been determined for JCV. Delivery of polyomaviruses to the ER is a critical step in infection. In the ER, both mPyV and SV40 interact with components of the host protein folding and quality control machinery for productive infection. It has been demonstrated that mPyV interacts with the protein disulfide isomerase (PDI) family of proteins PDI, ERP57, ERP72, and ERP29 (297, 521). The 72 VP1 pentamers that associate to form the viral capsid are stabilized by interactions between the C-terminal residues of neighboring pentamers, disulfide bonds formed between neighboring pentamers, and calcium ions coordinated by the viral capsid (276). Interaction with the PDI family of proteins, which can isomerize and disrupt disulfide bonds, is able to disrupt interpentameric binding and capsid stability. ERp29 contains only a single catalytic site yet retains its isomerization ability (187). ERp29 interacts with mPyV, and this interaction results in exposure of the C-terminal domain of VP1 (297, 401). Additionally, the previously internalized minor capsid proteins, VP2 and VP3, are exposed as a result of interaction with ERp29. This results in an altered virion that is capable of binding and penetrating lipid bilayers (297, 400). Since the exposure of the C-terminal domain of VP1 requires disulfide bond disruption, additional PDI proteins appear to be necessary for infection by mPyV. Recently, PDI and ERp57 have been identified as acting in a cooperative fashion with ERp29 to reduce and isomerize the mPyV interpentameric disulfide bonds (521).
In related work, similar ER-associated factors were found to play critical roles in SV40 disassembly (443). Inhibition of endogenous cellular PDI and ERp57 by small interfering RNA (siRNA) reduced SV40 infectivity. Unlike the case for mPyV, knockdown of ERp29 had no effect on SV40 infection. Additionally, in vitro interaction between ERp57 and SV40 uncoupled the 12 five-coordinated VP1 pentamers of SV40. It is likely that in the high-calcium environment of the ER, these pentamers remained attached to the virion. However, upon exit from the ER, the low-calcium environment of the cytosol allows for release of these pentamers.
A second critical step in the ER is the translocation of the virion across the ER membrane. As a result of recent studies, it appears likely that polyomaviruses utilize the ER-associated degradation (ERAD) pathway to traffic through a retrotranslocation pore and gain entry to the cytoplasm. Using short hairpin RNAs (shRNAs), Derlin-2 was identified as playing a critical role in mPyV infection. Derlin-2 is a protein involved in the transport of misfolded proteins out of the ER and into the cytoplasm to be degraded (277). Subsequently, Derlin-1 and SelL1, two proteins involved in the ERAD pathway, were found to be important in SV40 infection (443). These interactions are predicted to facilitate the release of a partially disassembled and destabilized virion into the cytosol. For SV40, it has recently been demonstrated that intact virions penetrate the ER membrane during retrotranslocation to the cytosol (210). This suggests that either the retrotranslocation pore used by SV40 is large enough to accommodate a 40-nm virion or the hydrophobic capsid residues exposed in the ER facilitate perforation of the ER membrane and allow an intact virion to enter the cytosol.
After retrotranslocation, the low calcium concentrations encountered by the virus in the cytosol further disrupt capsid stability. This destabilization is believed to result in shedding of a number of the previously attached viral pentamers (443). As a result, nuclear localization signals are exposed, resulting in transport of this partially disassembled virion to the nucleus through the nuclear pore (212, 226, 354, 548). It is likely that some combination of VP1, VP2, and VP3 remains associated with the viral genome during nuclear entry, as minichromosomes that contain the viral genome and nucleosomes do not efficiently enter the nucleus when microinjected into cells (354).
Thus, it is becoming increasingly clear that polyomaviruses utilize a complicated and novel pathway among viruses to ultimately gain access to the host cell nucleus. JCV binds sialic acids similarly to other polyomaviruses yet enters cells by clathrin-mediated endocytosis, suggesting that gangliosides may not play a role in JCV entry. Further experiments will be necessary to determine which receptor molecules contain the sialic acids to which JCV binds for productive infection. It will also be important to determine the similarities between JCV, mPyV, and SV40 in the later steps of virus entry and trafficking.
Unlike other polyomaviruses such as BKV and SV40, JCV shows a more restricted host cell range, which has made biochemical and molecular studies difficult. The viral T antigen interacts specifically with human DNA polymerase, restricting the host range in which JCV can replicate (142). Originally, JCV isolates were able to be grown only in human brain cells (540, 541). In order to facilitate biochemical studies of the virus, several other culture models were developed. SVG cells are T antigen-dependent, immortalized human fetal brain cells derived by transducing a heterogeneous culture of human fetal brain cells with a nonreplicating, origin-defective SV40 vector that expresses the SV40 T antigen (303). These cells allowed for the growth of JCV stocks and the study of viral gene expression and replication. Later, it was demonstrated that human fetal astrocytes in culture could support JCV replication, raising the possibility that JCV could replicate in cells other than oligodendrocytes (307).
Several other cell lines that support JCV replication have been established. A JCV-induced owl monkey glioblastoma cell line which spontaneously produced the Mad-4 variant of JCV was established in tissue culture (308). Transformed Rat2 cells were established to study JCV-induced transformation in culture (497), and the IMR-32 human neuroblastoma cell line was determined to support JCV replication, which made it useful for propagating virus (6). JCV produced by persistently infected IMR-32 cells became cell culture adapted for growth in these cells (368, 372). Several other cell lines were established by fusing primary human fetal astrocytes with the glioblastoma cell line U-87MG (441). Alternatively, a pseudovirus system was created using SV40 large T antigen-transformed COS-7 cells expressing JCV structural proteins, which produced virus-like particles lacking viral genomes (451).
More recently, cultures of human fetal brain progenitor-derived astrocytes (PDAs) or progenitor-derived neurons (PDNs) derived from human fetal neural progenitor cells have been used to study viral gene expression, replication, and growth in specific glial cell types (334). These cells allowed further study of the molecular regulation of JCV by cell-specific factors in a more pure population of cells.
Although pathology of JCV occurs in tissues of the central nervous system, seroepidemiological studies show that more than half the global population is either transiently or latently infected with JCV (378), although there is heterogeneity among populations, even in nonindustrialized areas (305). Additionally, the percentage of seropositive individuals increases with age (247), and 10 to 30% of people shed JCV in the urine (299). Because direct infection of brain tissue is unlikely and would be exceedingly rare, dissemination by some more common route is probable. Further inquiry led to the observation that cells of the immune system, including hematopoietic progenitor cells and B lymphocytes, as well as tonsillar stromal cells and Schwan cells are susceptible to JC virus (19, 341, 342, 345). Additionally, it was found that JCV may remain latent in immune cells of the bone marrow (198, 201, 322, 484).
Despite the limited range of species and cell types permissive for JCV replication, JCV is ultimately a very successful pathogen, as illustrated by its wide dissemination but rare pathogenesis. This success is attributable to the tight regulation of the viral life cycle. The differential ability of JCV to bind, enter, transcribe gene products, replicate, and ultimately produce more infectious viral particles in various types of human cells is essential to the regulation and life cycle of the virus. The following sections discuss these aspects of the viral life cycle, with emphasis on stages that regulate the virus' ability to complete its life cycle in various cells.
Similar to the case for all other polyomaviruses, the JCV genome is a closed circular supercoiled chromosome that is composed of “early” and “late” genes that are separated by the noncoding control region (NCCR), which contains the origin of replication (ORI), promoter, and enhancer elements (154) (Fig. 1 and and3).3). The early region is expressed de novo after infection and before DNA replication and is on the ORI-proximal side of the NCCR. Late genes are optimally expressed concurrently with or after DNA replication and are found on the ORI-distal side of the NCCR. The polyomavirus NCCRs are the most variable portions of the viral genome within a single virus as well as across genera of viruses (143, 151, 154, 267, 411, 446, 551).
The early-region sequences are transcribed counterclockwise from the NCCR and encode five proteins: the large T and small t antigens and splice variants called T′135, T′136, and T′165 (153, 395) (Fig. 1). The JCV large T antigen is a 688-amino-acid nonstructural, multifunctional protein that regulates viral early gene transcription and is thus autoregulatory. Large T antigen also regulates the switch from early to late viral transcription, as well as viral DNA replication. Large T antigen interacts with a number of cellular proteins to support transcription and replication of the viral genome, and upon unregulated overexpression, it induces cellular malignant transformation. The 172-amino-acid small t antigen shares the open reading frame (ORF) start site of nucleotide 5013 (339), and thus its 5′ end, with large T, but it is differentially spliced (154) and contains a translation termination signal at nucleotide 4495 (339), as opposed to position 2603 for large T antigen (154), which results in different 3′ ends of small t and large T antigens. Small t antigen and the other T antigen splice variants also perform multiple functions to drive infected cells toward S phase and may contribute to gene expression and PML progression to various extents in the presence of different underlying disorders (211).
The late region is transcribed clockwise from the opposite strand of the genome, and is composed of the coding sequences for four proteins (154) (Fig. 1). The smallest protein, agnoprotein, has an open reading frame that codes for 71 amino acids, beginning at nucleotide 277 and terminating at nucleotide 492. The agnoprotein is not well understood but has been proposed to be a viroporin (478), as well as to interact with large T antigen to decrease viral DNA replication. It has also been demonstrated to interact with transcriptional activators and repressors to control gene activation as well as influence DNA repair pathways (101, 238, 437). The other three open reading frames code for the viral structural proteins VP1, VP2, and VP3. The open reading frames for the structural proteins overlap. The 354-amino-acid major capsid protein VP1 is found at the 3′ end of the late region (nucleotides 1469 to 2533), is the major structural protein, and functions in cellular binding and entry. The ORFs for the minor capsid proteins, VP2 and VP3, are found between the 3′ terminus of the agnoprotein ORF and the 5′ end of VP1. VP2 is a 344-amino-acid protein with an ORF between nucleotides 526 and 1560. VP3 is composed of the C-terminal 225 amino acids of VP2, with its coding region starting at nucleotide 883 and sharing the 3′ terminus of VP2 (154, 301).
The NCCR lies between the early and late coding sequences and contains the origin of replication (152, 154, 339). It contains extensive regions of homology with the polyomaviruses SV40 and BK virus, including two of the three T antigen binding sites found in SV40 (152), as well as a unique non-B DNA tertiary structure (15). The NCCR is thought to be the main determinant of cell type specificity and is composed of fairly well-conserved flanking regions that border the transcription start sites of the early and late coding regions, as well as a central region containing numerous transcription factor binding sites. The early-proximal side of the NCCR contains the preorigin and origin of replication. The NCCR varies greatly between isolates from PML patients. In addition, a sequence, known as archetype or CY, has been isolated from urine specimens from both PML patients and healthy people but is rarely found in PML brain tissue. The original sequence isolated from a PML patient is known as Mad-1, as it was isolated at the University of Wisconsin—Madison. Subsequent isolates have been named with various designations, often depending on their associated immunosuppression or locations of isolation, with Madison isolates being known as Mad-1, Mad-2, and so forth.
The NCCR from the original isolate of JCV, Mad-1, contains an enhancer element that exists as a 98-bp direct tandem repeat and therefore contains duplicate TATA boxes, which can position mRNA start sites (165, 167), as well as multiple transcription factor binding sites (152). The tandem repeat structure of the Mad-1 NCCR variant has been termed the “prototype” JCV NCCR sequence and is composed of three blocks of sequence, named “a,” “c,” and “e,” with the TATA box found in “a” (Fig. 3A). Many studies have demonstrated that the TATA box(es) contained in the 98-bp direct repeat structure is essential for transcription of early and late viral genes (95, 230, 231, 234, 253, 503). Numerous NCCR variants containing a tandem repeat-like structure have been isolated from tissues of patients with PML (317). Although Mad-1 was the first isolated JCV NCCR variant, it has subsequently been shown that many JCV isolates from PML patients do not contain the second TATA box and that it may be dispensable for viral replication (293, 317). A naturally occurring variant of the NCCR, found in the CY or “archetype” JCV sequence, is composed of a single copy of the 98-bp repeat of a-c-e, with 23-bp (“b”) and 66-bp (“d”) sequence blocks between “a,” “c,” and “e” to yield an a-b-c-d-e structure (Fig. 3C). Archetype is rarely associated with PML tissue (554). The consistent isolation of tandem repeat-like NCCR sequences including the 98-bp tandem repeat in tissues obtained from PML patients strongly suggests the importance of this structure in viral pathogenesis (154, 214, 317, 322, 510). The prototype and prototype-like sequence variants are generally found in PML tissue, while kidney- and urine-derived NCCR sequences are normally homologous to archetype.
Comparison of the prototype NCCR with the archetype NCCR allowed identification of six blocks of sequence, “a” through “f” (24), and illustrated that archetype NCCR contains all of the nucleotide elements present in repeat structures as depicted in Fig. 3. Based on this comparison (24) and suggestions from BK virus studies (144, 427), it has been proposed that all JCV isolates contain NCCRs that are derivatives of the archetype sequence (144, 205, 554). Despite the presence of functional protein-coding regions and an origin of replication, archetypal NCCRs do not support robust growth in culture (96) and are isolated almost exclusively from the kidneys and urine (144, 286, 554). A mechanism for derivation of prototype sequence from archetype and dissemination of archetype in the host have yet to be demonstrated, although the prevailing model holds that archetype-like sequences are transmitted from person to person and then undergo deletions and duplications within the infected host, leading to PML-type NCCR sequences, which traffic to the brain. This “rearrangement” of the NCCR may take place in lymphoid cells like B cells, since they possess the Rag1 and Rag2 enzymes for immunoglobulin gene rearrangements.
Several other variants of interest have been identified. The Mad-4 variant is frequently used as a lab strain. It is identical to Mad-1 except that the NCCR contains a 19-bp deletion that eliminates the second, late-proximal TATA box. Mad-4 has a high incidence of oncogenesis in rodents (381). The coding regions of the Mad-4 isolate are identical to those of Mad-1 (317), so differences seen in cell culture models are directly attributable to differences in the NCCR. Additionally, an owl monkey inoculated intracranially with a donor tumor cell suspension containing Mad-4 developed an astrocytoma that was cultured and spontaneously secreted infectious JCV into the culture medium. After serial passage in culture, this virus also showed differences in the sequence and function of large T antigen (308).
In many PML patients, there is a predominant genotype of the NCCR, which probably coexists with minor subtypes with variations. The NCCR varies between patients, as well. Most of these variants contain a repeat, with both deletions and insertions compared to archetype and prototype NCCRs. For example, one early isolate, Mad-8 (Fig. 3C), is more typical of NCCR variants found in PML patients than Mad-1. This variant contains a repeat structure similar to that of the prototype, with one large deletion and one insertion, as well as several smaller insertions and base changes (316, 317). Examples of rearranged NCCRs can be found in many studies that have sequenced, annotated, and aligned multiple NCCRs from PML patients (413, 485, 552, 557).
Additional support for the role of the repeat NCCR structure in pathogenesis has been the identification of prototype-like sequences in cells suspected of harboring latent virus, including lymphocytes isolated from peripheral blood (121, 341, 444, 494) and the bone marrow (201, 322, 484). Interestingly, the B cell development from CD34+ progenitor to mature plasma cells capable of secreting IgG requires the expression of genes that support and carry out recombination (467). It may be that low levels of JCV genome in latently infected lymphoid cells are subject to rearrangements, which is supported by the environment in the developing B cell.
Regardless of how the repeat NCCR variants are generated, this form of JCV is the pathogenic form that has been repeatedly isolated from PML patient tissues. Because the prototype NCCR contains the repeat structure, it contains significantly more binding sites for the transcription factors essential to viral gene expression. In particular, the archetype sequence does not contain the Oct-6/tst-1/SCIP sites present on the border between regions “c” and “e” (Fig. 3) (510). Additionally, the lack of neighboring “a” and “c” regions eliminates Spi-B binding sites, which are important for early viral gene expression (314). The lack of binding sites for brain-specific transcription factors may be what abrogates the ability of archetype virus to cause disease in the brain. The lack of repeats of region C in archetype also leads to a reduced number of NFI binding sites, which allow a family of transcription factors to bind the JCV genome and are essential for fully activating viral transcription in the brain and cells of the lymphoid system. These results indicate the importance of selective repeated binding sites for the cellular transcription factors involved in activating viral gene expression.
Although Mad-1 was the first sequenced NCCR isolated from a PML patient and is referred to as the prototype, it appears to be a somewhat atypical isolate. The tandem repeat sequences serve as a good reference point, however, so newly isolated JCV sequences are generally compared to the Mad-1 variant. In order to classify newly sequenced viral NCCRs, a compass-like classification scheme that organizes NCCR sequences into four distinct variant types was developed (213). Variant type I NCCRs contain no inserts in the a-c-e organization of the NCCR and can be divided into variant type IS (singular a-c-e) and variant type IR (a-c-e repeat with no inserts) (e.g., Mad-1 and Mad-4). Variant type II NCCRs contain inserts into the a-c-e sequence and can be classified as variant type IIS (a-b-c-d-e, or archetype-like) or variant type IIR (containing inserts and repeats) (e.g., Mad-7 and Mad-8) (Table 1).
Prior to the system defining groups of JCV variants by NCCR architecture, several viral typing systems were developed. Early systems used restriction fragment length polymorphism typing (553), which can classify genotypes into 3 superclasses, A, B, and C (208), but with the increasing use of DNA sequencing, types were defined by genetic sequence. A region of 610 bp covering the 3′ end of T antigen, the intergenic region, and the 3′ end of VP1 has been used (1, 25, 205). More recently, a definitive coding sequence typing system was developed by coding region polymorphisms of 100 full-length JCV sequences, using predicted amino acid sequences of all the coding regions to define types. Using this system, 7 JCV types were identified, numbered 1 through 8 (type 5 was found to be a minor member of type 3 ), each with multiple subtypes (90). This typing system has determined the consensus sequence for all viral proteins and the consensus mutations associated with each subtype.
The different types of JCV are associated with populations of various descent (90) and have been used to map population movements (4, 208, 429, 473) as well as for other, diverse purposes, such as determining drinking water pollution levels (183) and identifying cadavers (206, 207). It has been hypothesized that type 6 is the original JCV type and that JCV coevolved with human populations. JCV split as humans migrated out of Africa, with one type moving toward Eurasia and the other type only to Europe (384). Type 1 and type 4 are generally associated with Europeans and European-Americans, while type 2A is found generally in Asians and Native American populations. Types 3 and 6 are isolated primarily from Africans and African-Americans. Types 2D and 7C are found among both Asians and South Asians (91, 550). Types 2E, 8A, and 8B are found in Western Pacific populations (550). Interestingly, type 8A is found only in populations of Papua New Guinea (218). JCV subtype 2B, which is more often found among Asians and Eurasians, has been associated with increased incidence of PML (2, 90), while type 4 has been associated with lower disease risk (124) (Table 2).
Most of the studies of viral subtype were performed on JCV DNA from urine specimens from infected yet healthy individuals. More recently, investigation of changes in viral DNA coding regions from brain biopsy specimens and CSF specimens from PML patients has been undertaken. Alterations in structural protein amino acid sequences could lead to enhanced viral entry and thus contribute to PML or, alternatively, could potentially cause distinct diseases. Evidence points to the association of mutations of certain amino acid residues in the region of the sialic acid binding sites and surface loops of VP1 with PML and that these mutations occur within the patient after initial infection with JCV (2, 148, 172, 286, 477, 556). In at least one case a frameshift mutation in the VP1 gene was found to be associated with JC virus granule cell neuronopathy (93). In some animal models, changes in VP1 and T antigen amino acid sequences led to an increased incidence of tumors (123, 147, 308).
In many studies, most of the population appears to be infected, either transiently or latently, with JCV. Epidemiological study of JCV depends on detection methods, and these methods have varied somewhat from study to study. Ten to 30% of adults excrete JCV in the urine, and PCR detection for viral DNA allows for accurate and sensitive detection of individuals with actively replicating virus. Virus has also been detected by PCR in stool samples and is prevalent in sewage and rivers worldwide (5, 183, 331, 505, 506). This has led to the hypothesis that transmission may occur from ingestion of nonsterile water. Transplacental transmission is unlikely (43), but transmission from parent to child can occur (42, 555).
Not all people infected with JCV excrete virus in the urine; therefore, other methods of determining infection rates have been utilized. Seropositivity for JCV was first determined using hemagglutination inhibition assays, but this has since been replaced by enzyme-linked immunosorbent assays (ELISAs), using recombinant VLPs (182, 290). More recently, a two-step anti-JCV antibody assay has been developed in an attempt to stratify disease risk by antibody presence. In addition to a standard ELISA using Mad-1-derived VLPs, samples are adsorbed with JCV VP1 and retested by ELISA. The percentage of reduction of ELISA detection indicates a seropositive sample, reducing false positives while maintaining specificity (171). In order to identify risk factors for PML, further research into epidemiology and subtypes of JCV, as well as continued study of transcriptional control of the virus, is required.
JCV has a limited host replication range. There are two identified blocks to replication in nonpermissive cell types: early gene transcription and DNA replication. Activated late gene transcription is initiated only after DNA replication begins, although DNA replication is not required for late gene activity in vitro (52, 229). A number of host factors are required and may contribute to JCV early transcription. In the presence of T antigen isoforms, T antigen, the host DNA polymerase, and a number of other cellular proteins participate in JCV DNA replication.
Nuclear domain 10 (ND10) bodies (also called PML nuclear bodies, after the promyelocytic leukemia protein, which makes up the scaffold of this structure, or PODS, for PML oncogenic domains) are discrete nuclear loci characterized by accumulation of the promyelocytic leukemia protein, as well as Daxx, SP100, and a number of other cellular proteins (359). ND10 bodies are the nuclear substructures at which JCV DNA replication and capsid assembly occur (454). ND10 bodies are thought to regulate a number of cellular processes, as well as participate in the life cycles of numerous viruses (139). Like for many DNA viruses, the JCV genome can be detected at ND10 bodies after infection (452). ND10 bodies show a complex biology with regard to viral gene transcription and DNA replication (452). Numerous studies indicate that they play a role in cellular defense against DNA viruses, but in a number of cases, ND10 bodies seem to be positive regulators of viral replication (139). Multiple viruses, including polyomaviruses, disrupt or reorganize the ND10 bodies during the course of infection (215). Many viruses, including JCV, localize at, or in the vicinity of, ND10 bodies (139, 222, 452–454). During SV40 and BK virus infection, DNA elements have been found to localize in the vicinity of ND10 (222), and this localization is required for efficient DNA replication but not for transcriptional activation of SV40 (489).
The localization to ND10 bodies may promote JCV DNA replication in a number of ways. A number of transcriptional activating and DNA replication proteins can be found at ND10 bodies, including CBP/p300 and proliferating cell nuclear antigen (PCNA) (453). Thus, it is likely that JCV uses the ND10 bodies as a scaffold for DNA replication and viral assembly during early stages of infection, while late in infection, JCV causes the disruption of these replication sites (453) but does not reduce the total amount of the ND10 scaffold promyelocytic leukemia protein. Interestingly, T antigen, which is absolutely required for viral replication, also localized to the ND10 bodies.
Functional T antigen is required for JCV replication. This was confirmed by the observation that JCV containing mutations in the T antigen-coding region cannot commence a lytic infection. Both the SV40 and BKV T antigens can bind to the JCV origin and initiate a lytic infection (86). The SV40 T antigen has greater DNA binding activity and is more efficient in directing replication than the JCV T antigen (44, 86, 294). The SV40 and JCV T antigens are able to bind to the origin of replication in two of the three GAGGC sites (152). Sequences on the late side of second T antigen binding site have been shown to be extremely important to replication. These sequences adopt an unusual tertiary structure (15; reviewed in reference 301), and this structure may be required for efficient T antigen-directed viral DNA replication.
The kinetics of JCV replication indicate a slow process. Even in susceptible cells in which T antigen is already present, DNA replication is undetectable for several days (303). DNA replication can be detected at 3 to 5 days postinfection in susceptible cell types and continues for several weeks (142, 233).
In addition to the cell type restriction for JCV growth at the level of early transcription, the viral life cycle is also restricted by cell type differences in DNA replication. JCV can replicate in immortalized primate cell lines expressing the SV40 T antigen (185, 303), as well as in primate cell lines expressing the HIV transcriptional activating protein, tat (369, 370). However, rodent cell lines immortalized with JCV T antigen, monkey cells, or nonglial human cells that do not express T antigen cannot sustain efficient viral replication. JCV replication could occur in the monkey and human immortalized cells in the presence of T antigen, but in the rodent cells, even T antigen expression could not stimulate replication. Thus, it appears that transcription is regulated by cell-specific factors, while the restriction of DNA replication is most likely regulated by species-specific factors. These species-specific factors, which may be a component or components of the DNA polymerase (142), allow JCV DNA replication only in primates.
Viral DNA replication proceeds as early viral proteins accumulate. Large T antigen binds preferentially to site II, located in the viral DNA replication origin closest to first TATA box (see Fig. 3) in the NCCR (47), but it also binds cellular DNA (375). When large T antigen binds JCV DNA, it promotes the YB-1/Purα switch to viral late transcription. Large T antigen also interacts with host cell replication machinery to directly initiate replication.
Replication of JCV DNA has not been as well studied as that of SV40 DNA but is likely to be similar. Like in JCV, the SV40 genome is a closed circular supercoiled DNA molecule. To initiate DNA replication, large T antigen forms a double hexamer and acts as a helicase and complexes with topoisomerase I, DNA polymerase α, and replication protein A (RPA) (60, 141, 360). Large T antigen also contributes to elongation of the DNA chain by its interaction with DNA polymerase δ, proliferating cell nuclear antigen (PCNA), and replication factor C (273, 501, 529). Replication proceeds bidirectionally, similar to theta replication, and leads to two interlinked DNA circles, which are resolved through the action of topoisomerases I and II (reference 360 and references therein). It has been proposed that linear SV40 genomes can initiate rolling-circle replication, generating concatemers, and that this method of replication may explain some of the recombination seen in viral variants (105, 122, 459).
In order to promote an environment conducive to DNA replication, T antigen binds to cellular proteins and DNA to induce signals to drive quiescent cells toward S phase (64, 119, 237). Large T antigen has been demonstrated to exhibit numerous functions, including interaction with, and inhibition of, the retinoblastoma protein (pRb) (48, 490, 533) and p53 (110). Interaction with p53 also prevents apoptosis induced by checkpoint activation when cells aberrantly enter S phase (112). Additionally, large T antigen can promote viral replication in G2-arrested cells by inducing DNA damage response pathways, and this function was related to binding of cellular DNA (375).
Small t antigen has been less well studied, but has been shown to interact with the RB family of proteins, as well as protein phosphatase 2A (PP2A) (45). The interaction of small t antigen with PP2A appears to prevent the dephosphorylation of the late protein agnoprotein, and this allows for greater viral replication (440). Reduction in levels of small t antigen or PP2A results in reduction of DNA replication (45, 440). Inhibition of the phosphatase activity of PP2A also drives the cell toward S phase, thus promoting viral replication.
Three alternative splice variants of T antigen exist, i.e., T′135, T′136, and T′165, which share their N termini with large T antigen, and disruption of their donor splice site results in greatly reduced viral replication (498). Although much study remains to understand their functions, putative interactions with Rb family members have been identified, which also appear to drive cells into S phase (46).
While it is difficult to separate the effects of increased viral early gene expression from DNA replication, several proteins have been implicated in directly increasing viral DNA replication. The NFI family of proteins has been shown in cell-free DNA replication systems to increase DNA replication of adenovirus type 2 (50, 347) and the replication of SV40 in vivo (348). NFI proteins have also been extensively shown to modulate JCV replication in vivo (14, 235, 483). The isoform NFI-A, which is expressed in several nonpermissive cell types, has been shown to decrease viral late protein expression (409), while NFI-X (NFI-D) increases viral gene expression and is highly expressed in cells permissive to JCV replication (344). The cellular protein Purα is also likely to participate in viral DNA replication, as it can bind the origin of replication and has been shown to repress viral replication (71, 157). Additionally, there is evidence that the protein Sμbp-2 decreases viral DNA replication, while its smaller variant, GF-1, which encompasses only some of the helicase motifs of Sμbp-2, may increase viral replication (78).
It is likely that because of the repeated sequence and nonstandard secondary structure of the NCCR, recombination, deletions, and insertions occur during viral DNA replication. These recombinations, insertions, and deletions can explain the large variations of sequences of NCCRs derived from PML patients. The prototype Mad-1 NCCR contains two identical tandem repeats (two sets of a-c-e sequence blocks) followed by an f sequence block. Most PML-derived NCCR sequences contain some version of these repeats, with deletions and insertions in some cases. Archetype, or CY virus, which is found primarily in the kidneys and urine and often found in healthy subjects, does not contain repeats and generally has a regulatory region consisting of sequence blocks a-b-c-d-e-f. One hypothesis for viral transmission and evolution is that archetype-like virus is the circulating form and that deletions (generally of b and d sequence blocks) then occur, followed by duplication of remaining sequence. This then leads to a pathogenic form of the virus able to replicate efficiently in glial cells (24, 198). There is evidence that the archetype “d” region may be inhibitory to JCV growth in some cells so that its deletion allows productive infection of other cells (173).
Alternatively, since archetype virus is rarely found in tissue outside the kidney, it is possible that prototype-like viruses are transmitted and that sequences are deleted or duplicated through base mispairing and single-strand slippage or through various forms of DNA recombination. The “b” and “d” blocks of sequence can be found in the human genome (L. J. Marshall and E. O. Major, unpublished data). These could be incorporated through DNA “capture.” This hypothesis has been used to explain the generation of host-substituted SV40 variants (reference 459 and references therein) and may apply to JCV as well. Recombination of the JCV NCCR may also be explained by its interaction with cells of the immune system (see below).
The encapsidated JCV genome is closed, supercoiled, and circular and is bound by nucleosomes derived from the four core histones of the previous host cell, as determined for SV40 (12, 41, 302, 333, 518). Once the genome is delivered to the newly infected cell, it acquires the linker histone H1 and resembles cellular chromatin (302). In the nucleus, the JCV genome serves as a template for the host RNA polymerase II (pol II) transcriptional machinery. Transcription of the JCV early genes occurs in the absence of de novo protein synthesis and utilizes only host proteins. Much, if not the majority, of the cell type specificity of JCV within human cells occurs at the transcriptional level. Regulation of transcription is dependent on the sequence of the NCCR, as well as the availability of host transcription factors.
Mad-1 NCCR transcription factor binding sites include four Oct-6/tst-1/SCIP binding sites (528), two Purα binding sites (79), two YB-1 binding sites (79, 232), two LCP-1 binding sites (479), two GF-1 binding sites (78), four NFI binding sites (13, 14, 455, 483), and six Spi-B binding sites (314). The Mad-1 NCCR is composed of two 98-bp tandem repeats, each containing a TATA box (Fig. 3A). The transcription start sites for early and late transcripts have been mapped in several cell types and for several viral variants. In addition to specifying cellular permissiveness to JCV infection, viral NCCR sequence and host factor availability help to determine transcriptional start sites.
The 5′ termini of early mRNAs at 5 days postinfection of primary human fetal glial cells were mapped to nucleotides 122 to 125 (using the numbering system introduced by Frisque et al. ) by S1 nuclease analysis, which maps to within the late-proximal TATA box. This contrasts with the case for in vitro-transcribed RNA, which mapped to nucleotides 94 to 97 (231). In JCV-transformed hamster brain cells, the start site of major early viral RNA was mapped to nucleotides 5115 to 5124, approximately 25 bp downstream of the early-proximal TATA box, while a second, minor pair of early mRNA start sites was positioned approximately 25 bp downstream from the late-proximal promoter, by which they were probably positioned (310).
The Mad-4 NCCR has a deletion in the second repeat that eliminates the late-proximal TATA box and allowed for studies that determined the functional significance of multiple TATA boxes. The early mRNA start site of Mad-4 also mapped to nucleotides 5115 to 5124 in transformed hamster cells, but the minor start sites were eliminated, indicating that they were indeed positioned by the late-proximal TATA box (310).
Rodent cells are not permissive for JCV replication and are therefore transformed by JCV, so the transcriptional start sites may be different than in the natural human host. However, similar start sites were found in the human fetal glial cell line POJ, which constitutively expresses a functional JCV T antigen expressed by replication-defective JCV (309). In primary human fetal glial cells, at 3 to 5 days postinfection, the start sites of Mad-1 were found at nucleotides 5122 and 5082 by primer extension, whereas at 10 days postinfection, after DNA replication had begun, the early mRNA start sites shifted and were found at nucleotides 5012, 5037, 5047, downstream from the early-proximal TATA box, and at nucleotide 35, which is between the first and second TATA boxes (233). In another study utilizing both S1 nuclease and primer extension techniques in primary human fetal glial cells, early mRNA start sites were found for Mad-1 approximately 25 bp downstream of both TATA boxes, at nucleotides 89 to 92 and 5115 to 5125 (95). In that study, as in previous studies, the Mad-4 early start site at nucleotides 5115 to 5125 remained, but the site positioned in Mad-1 by the deleted TATA box was eliminated.
These data indicated that both TATA boxes were functional for early transcription and that they may vary in importance depending on the cell type infected. Although the second TATA box directs minor mRNA start site positioning, the major early mRNA start sites are downstream from the first TATA sequence, and the second TATA box may be dispensable. Many of the NCCR variants found in PML patients do not contain the second TATA box (90, 173, 317, 413). Recently, it has been shown that the Spi-B transcription factor binding sites in the second repeat can compensate for the loss of the second TATA box (314). Additionally, NCCR variants lacking repeat sequences show greatly reduced early transcriptional activity in comparison to both Mad-1 and Mad-4 (173). Thus, sequences and transcriptional activator binding sites that are important for early gene transcription are present throughout the viral NCCR and contribute to viral early transcription, even in the absence of a second TATA box.
In human fetal glial cells, at 17 to 19 days postinfection, several major and minor transcriptional start sites for Mad-1 late mRNA were found to span a large region of the viral NCCR, covering approximately 250 bp between nucleotides 5114 and 242, which are on either side of the repeats (95, 230). These start sites appear to be positioned not by either TATA box but rather by the sequence TACCTA, which was approximately 30 nucleotides upstream from the minor start site at nucleotides 90 to 98, as well as the major start site at nucleotides 198 to 203 (230). The TACCTA sequence can function as a surrogate TATA box, as has been shown in SV40 (53, 355).
Host transcription factor availability is the determining factor in both the start sites for early transcription, as well as the quantity of T antigen produced. Although NCCRs containing repeats (such as Mad-1 or Mad-4) and variants isolated from PML strains have greater transcriptional activity in PDA cells (173), both the archetype and various PML isolates show increased transcriptional activity in glial cells rather than cells of nonglial origin (23, 464). Once T antigen is present, the difference in replication fitness between different variants of JCV becomes much less apparent (23). Therefore, the ability of the virus to transcribe the early side of its genome is a major determinant of cell type specificity of the virus and has been the focus of much of the research on viral transcription.
Unlike other human DNA-containing viruses, such as herpesviruses, JCV does not bring transcriptional activating proteins into a newly infected cell. Thus, early transcription is directed entirely by host cell factors. The JCV NCCR contains binding sites for a number of transcription factors and transcriptional repressors. As shown in Fig. 3, the proteins NFI-X (344, 410), DDX-1 (475, 476), LCP-1 (479), HIF-1α (390), BAG-1 (118), NFAT4 (311), NF-κB (405, 435), GF-1 (78), SP1 (190, 191, 242), and Spi-B (314) have all been proposed to bind to certain variants of the JCV NCCR and activate early transcription in various cell types, while NFI-A (409), c-jun (240, 410, 432), c-fos (240), SF2/ASF (439), and C/EBPβ (425) have been proposed to repress early transcription. Of these, the most well-studied JCV-transcription factor interactions have been with members of the NFI family.
The NFI family of cellular DNA binding proteins is critical to JCV transcription and replication (344, 409, 410). These proteins were first identified as part of the minimal set of proteins required for in vitro replication of adenovirus DNA (106, 176, 351–353). Three NFI binding sites have been demonstrated in the JCV regulatory region (NCCR) by sequence homology and DNase protection assays (14, 483), and one potential, uninvestigated site exists, as identified by sequence homology. The confirmed sites have been alternatively labeled in order from the early side of the genome as NFI-A and NFI-B (285) or from the late side of the genome as NFI I, II, and III. In order to avoid misunderstanding, the second convention is less confusing, as four NFI genes exist, NFI-A, -B, -C, and -X (alternatively called -D) (176). The two identical sites in the 98-bp palindromic repeat element (NFI II/III) were shown to be more important for early gene transcription than the unique third element on the late side of the NCCR (NFI I) (260, 455). These sites were also shown to be more important for late gene expression (259). Additionally, these sites appear to contribute to the cell type specificity of JCV, as they are bound by different proteins in different cell types (13, 14, 220, 259, 260, 409, 410, 483).
DNase I footprinting experiments demonstrated that various tissues and cell types contained different factors that bound to the JCV NCCR at dissimilar sites (482). This observation was explained when the four NFI genes were discovered, each with multiple splice variants (176). The dimerization, DNA binding, and DNA replication domains of NFI proteins are found in the N terminus and are separable from the transcriptional activating domains (176). All four NFI genes share homology at the N terminus and differ at the C terminus, which is responsible for transactivation and repression activity (176). NFI proteins can homo- and heterodimerize and compete for identical binding sites. The character of the NFI dimer may influence transcriptional activity. This could explain why overexpression of NFI-X confers the ability to support increased viral activity in cell types normally nonpermissive for JCV (344), while overexpression of NFI-A decreases the ability of permissive cell types to support JCV infection (409). Additionally, NFI proteins also regulate transcription from a number of genes important in cells of neuronal origin (324).
Members of the activating protein 1 (AP-1) family play an important regulatory role in JC virus transcriptional activation (13, 240, 410, 432). NFI binding to and activation of JCV are reduced by the presence of c-jun (13, 410). This appears to be due to the overlapping AP-1 and NFI binding sites in the JCV NCCR, implying that c-jun physically blocks NFI-induced activation. Interestingly, this juxtaposition of AP and NFI binding sites occurs at numerous genes associated with central nervous system cells, such as those for glial fibrillary acidic protein (GFAP), a cytoskeletal marker for astrocytes, and myelin basic protein (MBP), a component of myelin produced by oligodendrocytes (13). Thus, it is likely that activation of JCV is controlled by access to DNA binding proteins available in cells of the nervous system and that a common regulatory mechanism of JCV and these genes may exist. Further research will yield a greater understanding of the interplay between these DNA binding factors, the JCV regulatory region, and genes important for neuronal and glial differentiation.
Both NFI and AP-1 family members interact with large T antigen. NFI has been shown in several studies to increase early and late gene expression in a T antigen-dependent manner (14, 259, 260, 285), as well as to contribute to increased viral replication (344, 409, 410, 465). The AP-1 members c-jun and c-fos have been shown to physically interact with large T antigen and suppress its activation of early genes, as well as viral DNA replication (240). Thus, AP-1 family members and NFI family members appear to make up an antagonistic switch system for JCV gene expression and DNA replication.
This system may be similar to a better-characterized switch in the JCV life cycle. The cellular proteins YB-1 and Purα interact with the viral large T antigen to regulate the switch from early to late gene expression (77, 79, 232, 434, 436, 437). Purα is a strong activator of early gene expression and binds the viral lytic control element (LCE) (77, 79). As T antigen accumulates, it facilitates binding of YB-1 to the LCE, and together YB-1 and T antigen increase the displacement of Purα from the viral promoter. YB-1 and T antigen stimulate late gene expression (77, 79, 232, 436), and thus Purα, YB-1, and T antigen work as a genetic switch to shift gene expression from viral early to late genes.
Activated late gene expression requires T antigen and occurs concurrently with DNA replication but, at least in the case of SV40, does not require DNA replication to proceed (229). The large T antigen ORI binding function is not necessary for activation of late transcription of SV40 (229). Instead, T antigen promotes late transcription by interacting with components of the basal transcription machinery, including TATA binding protein (TBP), TBP-associated factors (TAFs), and transcription factors, including Sp1 (239). T antigen may also function directly as a TAF (92).
A number of DNA binding proteins have been implicated in regulation of viral late transcription. Egr-1 (424), HIF-1α (390), GF-1/Sμbp-2 (78), BAG-1 (118), and NFAT4 (311) have been shown to activate late gene transcription, while C/EBPβ (425) and GBP-i (402) appear to function as transcriptional repressors. Subunits of NF-κB have been shown to increase late gene expression (328, 435) and can increase viral expression in response to tumor necrosis factor alpha (TNF-α) stimulation (539). Subunits of NF-κB also interact with YB-1 (403), which stimulates the switch from early to late gene expression.
Additionally, although the functions of the late viral protein agnoprotein remain to be elucidated, it has been posited to interact with YB-1 and may modulate its activity (438). Agnoprotein may also interact with large T antigen to reduce T antigen enhancement of late gene expression (433).
The best-characterized cellular protein known to stimulate viral late gene expression is Tst-1, also referred to as Oct-6 or SCIP. Oct-6/Tst-1/SCIP is a POU domain protein known to function in neuronal development (209, 266, 527), thus potentially contributing to the cell type specificity of JCV transcription. Oct-6/Tst-1/SCIP overexpression increases both early and late gene expression (528). Oct-6/Tst-1/SCIP binds to large and small T antigens in vitro and can synergistically increase both early and late gene expression in the presence of large and small T antigens (416).
JC virus displays a complex interaction with cells of the immune system. Because of the lack of an animal model coupled with virtually undetectable viral levels outside the kidney, the study of latent sites of infection that are affected by altered immune responses has been difficult. Also, determining the immune response to JCV infection has been generally confined to those patients with altered immune systems in whom JCV replicates and causes PML. Additionally, certain cells of the immune system are susceptible to JCV and play a critical role in the viral life cycle as well as the pathology of PML.
Although JC virus has been known to be the causative agent of PML since the 1970s, the routes of initial viral transmission and subsequent dissemination to the brain remain to be fully elucidated. Without the provision of exogenous T antigen, JCV was known to replicate only in human cells of glial origin in culture. A hematogenous route of infection of the CNS seemed likely after JCV was discovered to interact with B cells in the brain, periphery, tonsils, and bone marrow and to replicate at low levels in B cells (22, 201, 300, 301, 421, 494).
Several years later, it was shown that JCV could infect tonsillar stromal cells and hematopoietic progenitor cells, as well as primary B cells (341, 345). In addition, evidence was provided that one of the initial sites of infection could be stromal cells of the tonsils (342). These discoveries led to a model in which primary infection most likely occurs in either stromal or immune cells of the upper respiratory system, either through respiratory inhalation or orally through ingestion of virus-contaminated material, possibly from excreted virus in the urine (37). The virus is then trafficked to the bone marrow and kidneys by infected lymphocytes, where it can persist for the life of the host. Upon immunosuppression or a change in immune cell trafficking, the virus mobilizes from the bone marrow and crosses the blood-brain barrier, either alone or in conjunction with a B cell. Once oligodendrocytes become infected, lytic infection begins (201). It is also possible that virus in the kidney replicates at high copy levels, escapes into the peripheral circulation, is taken up in lymphoid tissues such as the bone marrow, and then undergoes rearrangements of the NCCR.
Alternatively, the brain or kidney may serve as a site of latency (104). JCV DNA, but generally not protein, has been found in the brains of both healthy and immunocompromised patients without PML or other neurological disorders (31, 107, 386, 485). This suggests that JCV has access to the CNS before disease onset and may travel to, and nonproductively infect, the brains of some immunocompetent individuals. It is possible that after initial infection by JCV and viral dissemination, possibly through hematopoietic precursors or B cells, the virus reaches glial cells of the brain, where it remains latent. This observation leads to model in which JCV traffics to the brain and remains latent in the CNS. The virus would remain latent unless changes in the NCCR, which may occur before or during latency in the brain, and available binding factors occurred in the presence of immunosuppression. However, if these events occur, then upon a reduction in immune surveillance and control due to compromise or modulation of the immune system, JCV may reactivate in situ and cause PML. This pathway, however, does not address the very low incidence of PML in allograft recipients who are immunosuppressed for substantial periods for graft protection. In both models of viral latency, although the site of latency may differ, similar events must occur for progression to PML.
Approximately 60 to 80% of humans produce antibodies against JCV, indicating that the majority of the population has been exposed to the virus (247). These rates vary greatly among populations and age groups (305). Additionally, at any given time, approximately one-fifth of the population sheds JCV in urine (299). Only a very small fraction of these individuals become ill, however, and this occurs only in the presence of underlying changes to the immune system. Thus, in the majority of cases, JCV infection is controlled by the healthy immune system.
Several observations indicate that the cellular immune response plays a vital role in control of the virus. PML is an AIDS-defining illness, occurring in 3 to 5% of HIV-infected individuals (299). Before the advent of the AIDS pandemic, PML was extremely rare, indicating that a reduction of CD4+ T cells due to HIV infection leads to lack of immune control of JCV. Additionally, non-HIV-associated CD4+ T cell reduction due to idiopathic CD4+ T lymphocytopenia has been associated with a number of cases of PML (179, 396). Other studies have implicated an impairment of T cell responses in the development of PML (526), while a cytotoxic T lymphocyte (CTL) response has been associated with greater control of JCV and longer PML survival rates (129, 249, 250, 278, 321). The use of HAART has reduced the rate of PML in HIV-infected individuals, further indicating an important role for a cellular immune response to JCV in control of infection (10, 136).
In contrast, the humoral immune response has not been shown to control JCV infection (249, 526). In fact, the virus seems to have adapted to replicate and disseminate, possibly through B cells and their progenitors. JCV is known to remain intranuclear once assembled, which may allow viral escape from immune recognition.
In addition to serving as a potential site of viral latency, B cells may play an important role in the pathogenesis of PML. Since it has been posited that the viral genome recombines and/or rearranges during DNA replication, an attractive model is that these events occur in B cells, which can undergo V(D)J recombination. This hypothesis is bolstered by the observation that diverse viral NCCRs, including archetype-like and prototype-like NCCRs, have been found in the blood and bone marrow (214, 322, 484).
Recombination that results in prototype-like viral NCCRs is associated with increased viral activity in glial cells (173). JCV infection has also been shown to upregulate the DNA damage response (101, 102), and high antibody titers to JCV are associated with increased chromosomal damage in lymphocytes (269, 340). JCV infection of cells in culture as well as cultured lymphocytes also results in a high degree of chromosomal damage (356, 357). These damaged chromosomes are similar to those seen in SV40-infected cells and may be the source of some of the sequence blocks found in PML-associated viral NCCRs, such as in Mad-1. Thus, viral recombination may be explained by chromosomal damage induced by JCV in cells in which recombination and DNA repair mechanisms are active, as may be the case for SV40 (459). These changes may lead to acquisition of transcription factor binding sites in the NCCR that are important for pathogenesis. A recent example was described in patients receiving infliximab, where an archetype-like NCCR contained sequences that led to TATA box-associated Spi-B sites known to be important for viral replication, while JCV in the urine contained an archetype NCCR sequence (32). Additionally, as B cells mature, different transcription factors that play a role in increased viral proliferation are upregulated.
As early as the 1990s, it was recognized that there are DNA binding proteins that are common between B cells and glial cells (300, 421) but that do not exist in T cells (421). At least two factors shown to be important in JCV transcription and regulation, NFI-X and Spi-B, have been shown to be upregulated in B cells, glial cells, and hematopoietic progenitor cells in which JCV can replicate (314, 334, 344). Evidence that changes in transcription factors can affect viral transcription as B cells mature is increasing, particularly in light of the observation that natalizumab treatment upregulates factors involved in B cell differentiation, including Spi-B (281).
B cells may also carry JCV across the blood-brain barrier. Evidence for this is found in the fact that PML was first associated with B cell lymphoproliferative disorders (57, 198) and that natalizumab treatment-associated PML occurs concurrent with mobilization of lymphocytes from the bone marrow to the periphery (406). Additionally, HIV depletion of lymphocytes in the periphery may lead to mobilization of lymphocytes from the bone marrow to the periphery, and PML may be “unmasked” after the reconstitution of the immune system in the periphery by HAART treatment (457).
In addition to mobilization of B cells into the periphery, JCV must cross the blood-brain barrier to initiate infection of oligodendrocytes. B cells can carry JCV to the blood-brain barrier, where it may cross as free virus. JCV may also infect microvascular endothelial cells and thereby cross into the brain (75). Alternatively, B cells may carry JCV across the blood-brain barrier. For instance, in HIV infection, in which a relatively high percentage of infected patients develop PML, macrophage chemoattractant protein 1 (MCP-1) is upregulated, which increases the permeability of, as well as lymphocyte migration across, the blood-brain barrier (530). Infected B cells have been found in the CNS of a patient with PML (300). Additionally, infected B cells can transmit JCV to glial cells in culture (73).
The primary working hypothesis for development of PML is that at least four events must occur for latent JCV to cause lytic infection of the oligodendrocytes in the brain: (i) the host immune system must be compromised or altered, (ii) the viral NCCR must acquire changes that increase viral transcription and replication in both B cells and glial cells, (iii) DNA binding factors that bind to recombined NCCR sequence motifs must be present and/or upregulated in infected hematopoietic progenitor, B cells, and/or glial cells, and (iv) free virus or virus in B cells must cross the blood-brain barrier and be carried into the brain, where virus is passed to oligodendrocytes and lytic infection takes place. Once the virus is in the brain of the susceptible (immunocompromised) host, PML occurs. These events may occur in the bone marrow, in CD34+ lymphocyte precursors or B cells in the periphery, or in the brain.
Several immunomodulatory therapies have been associated with PML (66, 67, 146, 169, 245, 258, 262, 358, 463, 504). These therapies are promising for the treatment of a number of autoimmune conditions and lymphoproliferative disorders. PML has been identified as a serious adverse event associated with some of these therapies, which led the FDA to require a labeling warning. The known mechanism of action of each of these therapies also sheds light on mechanisms of immune control of JCV (for a more detailed description of the role of immunomodulatory therapies in the development of PML, see reference 299 and references therein).
Natalizumab is one of the biological therapies that carries a high risk of PML. It is a humanized monoclonal antibody for the treatment of relapsing-remitting forms of multiple sclerosis. It binds the α4 chain of the α4/β1 and β7 integrin dimer, also known as very late antigen-4 (VLA-4) (135). VLA-4 mediates cell migration and infiltration in immune signaling. VLA-4 binds to its ligand, the vascular cell adhesion molecule (VCAM), and participates in facilitation of extravasation of leukocytes through endothelial cells to the sites of inflammation (418). The α4 integrin can also dimerize with the β7 integrin, preventing T cell binding to mucosal addressin cell adhesion molecule 1 (MAdCAM-1) and extravasation into the gastrointestinal mucosa (299). Multiple sclerosis is characterized by chronic leukocyte infiltration into the brain, and natalizumab blocks this infiltration by preventing extravasation through cell adhesion molecule binding. Natalizumab treatment results in an increase in CD34+ progenitor cells in both the bone marrow and the blood (217). It also increases circulating pre-B and B cells in the periphery and prevents homing of CD34+ progenitor cells to the bone marrow and of pre-B cells to lymph node marginal zones (254, 289, 299). Natalizumab treatment also results in an increase of factors involved in B cell differentiation, including Spi-B, in the peripheral blood (281). As Spi-B has been shown to increase JCV transcription, this may be a mechanism for the high risk of PML in those treated with natalizumab (314). Recently, it has been shown that Spi-B is increased in CD34+ cells and B cells in natalizumab-treated patients (Marshall and Major, unpublished data). The risk of PML during natalizumab treatment rises as treatment progresses, and the true incidence of PML due to current immunomodulatory therapies remains to be determined, but it has been estimated to be approximately 3.85 per 1,000 patients treated with more than 24 infusions (available for prescribing physicians at https://medinfo.biogenidec.com).
Rituximab is an anti-CD20 humanized monoclonal antibody that fixes complement. Binding of CD20, expressed on B cells, results in downregulation of the B cell receptor and cytolytic apoptosis of CD20+ B cells (412). Administration of rituximab results in depletion of CD20+ B cells in the peripheral blood and CSF (299, 330). Rituximab treatment has been associated with severe viral infections (9) and JCV-induced PML (299). As with natalizumab, pre-B and B cells may be mobilized from the bone marrow and lymph nodes to replace depleted CD20+ B cells in the periphery, and there is an association with higher levels of CD34+ progenitors in the periphery (299).
Efalizumab is a humanized monoclonal antibody against CD11a, a subunit of the leukocyte function-associated antigen type 1 (LFA-1), a T lymphocyte adhesion molecule. LFA-1 binds intercellular adhesion molecule 1 (ICAM-1), which allows migration of T lymphocytes from circulation into sites of inflammation (271). Efalizumab also downmodulates expression of VLA-4 and results in T cell hyporesponsiveness (177). Efalizumab was voluntarily withdrawn from the market because of the occurrence of PML at an incidence of approximately 1 in 500.
Infliximab is a humanized monoclonal antibody against tumor necrosis factor alpha (TNF-α) (246). Infliximab also induces apoptosis in TNF-α-producing T cells (507, 508). It has been associated with an increase in infections or reactivation of latent infections (120). This is probably due to a reduction in cellular immunity due to the blockage of TNF-α and T cell reduction.
Mycophenolate mofetil is a small-molecule prodrug used to prevent rejection of organ transplants (283). It is metabolized by the liver to become mycophenolic acid, which blocks B and T cell proliferation by inhibiting IMP dehydrogenase and preventing purine synthesis (407). It is unclear how mycophenolic acid administration leads to PML.
It is likely that some of these therapies lead to PML due to a decrease in immune surveillance. Conversely, several of these therapies, notably natalizumab and rituximab, result in a decrease of mature B cells in the periphery and a subsequent mobilization of immature B cells from the bone marrow, potentially disseminating latent virus to the brain. Recombination of DNA in B cells also offers an attractive model for the changes in the viral NCCR that are necessary to increase pathogenicity and replicative efficiency of the virus in glial cells.
PML is an AIDS-defining illness and is the cause of death in 3 to 5% of AIDS patients (299). This rate of disease is significantly greater than that in patients with other underlying causes of immunosuppression. This may be due to several factors, including, but not limited to, duration and extent of immunosuppression, changes in cytokine secretion induced by HIV, viral interactions in coinfected cells, and increased blood-brain barrier permeability allowing for B cells infected by JCV to enter the brain (198).
As discussed above, CD4+ T cell number and specificity are major determinants of JCV infection of the brain and occurrence of PML. Additionally, CD8+ T cell responses specific to JCV are important to control of JCV (126, 250, 278, 279, 544). During chronic viral infections, CD4+ T cells are required to maintain a CD8+ T cell response (326). Dysfunction of B cells and increased circulation of B cells, which may favor JCV crossing of the blood-brain barrier, have also been observed during HIV infection (35). Thus, HIV infection seems to promote an immunological state that favors the onset of PML.
Additionally, studies have shown a potential synergistic role of HIV and JCV at the molecular level, an effect that is likely a cause of the high rate of PML in HIV-infected individuals. Interestingly, HIV and JCV may share an immune cell site of latency, as both JCV and HIV have been reported to be present in CD34+ bone marrow progenitor cells and may be reactivated upon differentiation to B lymphocytes (68, 198, 341, 345). JCV and HIV can also both infect astrocytes, although the fates of infection differ (87, 495, 514), and brain-derived progenitor cells (197, 268, 445). JCV and HIV have the potential to directly interact in multiple cell types.
The HIV tat protein has been shown to increase transcription from JCV (82–84, 97, 370, 415, 472, 481), while the JCV agnoprotein may cause a slight decrease in the replication of HIV (224). Archetype JCV, which normally cannot be efficiently propagated in cell culture, can replicate in cells expressing HIV tat (369, 370). tat has also been shown to be secreted from infected cells and internalized by uninfected cells, affecting cellular function (137, 138, 203, 491, 531) and thereby abrogating the necessity of JCV and HIV coinfection for the molecular interaction between these viruses. The uptake of HIV tat has been shown in oligodendrocytes, which supports the possibility of increased JCV transcription through the interactions of tat with various activating proteins (97, 133). The viruses also share requirements for transcription factors, including members of the NFI family (220, 344), and both viruses interact with the cellular protein Purα (77, 156, 532). tat interaction with Purα has been shown to increase late JCV transcription (252) and replication (98). Interestingly, the interaction between HIV tat and cellular Purα has been shown to play a role in DNA repair (525), which could potentially cause increased JCV rearrangements in coinfected cells, leading to an increased chance of JCV NCCR sequences associated with PML.
The transcription factor Sp1 may also play a role in the link between JCV and HIV. In one study, Sp1 sites found in the “b” segment of the archetype NCCR were deleted and TAR-homologous tat binding regions were duplicated in AIDS-associated PML, while in HIV-negative patients, the Sp1 sites were retained. This may be explained by adaption of the JCV NCCR to contain binding elements for HIV tat instead of the Sp1 binding site for more efficient activation in HIV-infected individuals (338). Conversely to the upregulation of JCV transcription by HIV, JCV agnoprotein associates with HIV tat in coinfected astrocytes and represses tat-mediated HIV transcription, in part by inhibiting tat interaction with Sp1 (224).
Additionally, HIV infection increases permeability of the blood-brain barrier (BBB) (23, 206). This may be a function of the viral tat protein, which can activate CCL2, leading to increased BBB permeability (38, 387, 530), and is associated with an increased incidence of PML (318). The increased permeability may contribute to JCV crossing of the BBB in infected B cells or as free virus. HIV infection of the brain also causes upregulation of cytokines which attract lymphocytes (34), as well as an increase in cell adhesion molecules which may facilitate crossing of JCV-infected cells (367). HIV proteins, such as tat and nef, can cause damage to astrocytes (275, 323), and direct infection of astrocytes by HIV may lead to neuronal damage (87, 495, 514). This damage may lead to increased inflammation and further infiltration by JCV-infected lymphocytes and may help facilitate onset of PML.
Treatment options for AIDS-associated PML are extremely limited (see reference 192 and references therein). There is no approved JCV-specific treatment and hence no treatment for PML. As discussed below, many drugs and biological molecules have been investigated for PML treatment, but the results in vivo have not been favorable. The sole treatment for PML in AIDS patients is highly active antiretroviral therapy (HAART), which has been shown to have no direct effect on JCV replication and thus is effective against JCV by limiting HIV replication and thus tat production, as well as allowing for increased immune function. Immune reconstitution in AIDS-associated PML has the potential to greatly increase the life spans of PML patients but carries with it the potential for severe and life-threatening side effect from immune reconstitution inflammatory syndrome (IRIS) (reference 219 and references therein). In some cases, PML has been “unmasked” by HAART, in that onset of PML occurred after the beginning of HAART (457). These observations suggest that virus may travel to the CSF following remobilization of lymphocytes during immune reconstitution.
More study is needed to determine the functional interplay between JCV and HIV, but it is clear that there is significant interaction between the viruses at the molecular, cellular, and immunological levels. Both viruses are neurotropic, infect cells of the immune system, cross the BBB, infect astrocytes, and cause destruction of oligodendrocytes, leading to a decrease in myelin (408).
Primary infection with JCV is asymptomatic and occurs in immunocompetent individuals early in childhood. Thus, the major risk factor for developing this infection is immune compromise. This includes patients with cancers that involve the lymphoid system, such as lymphomas, or patients who are on long-term immunosuppressive therapy for treatment of cancer or autoimmune diseases. Patients who undergo organ transplant are also at risk due to the need for chemotherapy to prevent organ rejection (325). Some elderly patients may also develop PML, presumably due to compromise in the cellular immune responses associated with the process of aging (164). The recent development of monoclonal antibodies that target different arms of the immune system has also been associated with PML. The majority of these cases are due to the use of natalizumab (anti-α4 integrin), which prevents T cell trafficking into the brain and is used for treatment of multiple sclerosis, and rituximab, which targets B cells (anti-CD20). However, cases have also been reported with alemtuzumab (anti-CD52), infliximab (anti-TNF), efalizumab (anti-CD11a), and ibritumomab (anti-CD20) (227). Immune compromise alone may not be sufficient to cause PML. It is possible that other host factors may also play a role. For example, in one study, sequencing of the p53 gene, exon 4, showed heterozygosity (Arg-Pro) at codon 72 in five of six PML patients (394).
Currently, a great deal of effort is focused on the stratification of those at risk for developing PML. Several factors that do not appear to be predictive of PML risk include viruria and viremia. Detection by quantitative PCR (qPCR) of JCV DNA in urine occurs at a higher rate in HIV-positive individuals than in the general population (272, 327), and while natalizumab treatment is associated with increased, but possibly transient, viruria and viremia (80, 422), 20 to 40% of the general population also demonstrate periodic excretion of JCV in the urine. The presence of JCV in the urine correlates with higher anti-JCV antibody levels, but viral copy number does not correlate with anti-JCV antibody levels (171). Based on these observations, the absence of urinary detection of JCV is not a reliable indicator of PML risk, although the presence of JCV DNA is a risk factor in that it confirms that JCV is present in the individual. Detectable JCV in the serum has not been correlated with progression to PML (515), although the risk associated with JCV detection in specific cellular compartments remains to be elucidated.
In natalizumab-treated patients who are dosed monthly, the greatest risk factors for development of PML are positive anti-JCV antibody status, previous immunosuppressant use, and treatment duration of greater than 24 to 36 infusions (225, 466, 515) (current information is at https://medinfo.biogenidec.com). No particular duration or type of prior immunosuppressant use has been identified as increasing the risk of PML. Because no evidence of residual immunosuppression existed when patients were started on natalizumab (225), persistent immunosuppression may not serve as a useful marker for higher risk of PML in natalizumab therapy. Current data do not allow for complete risk analysis of patients receiving greater than 36 infusions of natalizumab, but data will be available as more patients are treated for longer periods.
Current evaluation of anti-JCV antibody titers as a risk factor is problematic. Multiple assays are available, including hemagglutination (HA) inhibition assays (357), ELISAs using purified virus (182), ELISAs using virus-like particles produced in either insect or human cells (155, 516, 517), and a two-step assay developed by Biogen-IDEC (171). Additionally, assays currently lack a true positive control (such as humanized antibody against JCV VP1) and instead use pooled human sera of unknown status to provide a “positive” optical density (OD). These assays generally rely on the optical density of control sera, which are also of unknown antibody status, for a negative cutoff value, so no titration of levels or absolute titers can be determined. Assays also often use one dilution of serum, as opposed to serial dilutions of serum, increasing the possibility of false negatives. With such differences between assays, comparisons between tests and even laboratories performing the same assay are difficult.
Seronegative patients are in the group with lower risk for PML, so a test that determines the presence or absence of anti-JCV antibodies is currently of value, as long as it is consistent and has a low false-negative rate. The quantitative anti-JCV antibody titer may be of clinical value when deciding on courses of treatment, in that antibody titer may be of predictive value for PML risk. It will be important to standardize assays for anti-JCV antibody titer values to be useful to clinicians, and efforts are being made in this area (391). Physicians must rely on patient history, length of exposure to immunosuppressants or immunomodulatory drugs, and anti-JCV antibody status to guide treatment decisions. JCV DNA in the urine also demonstrates the presence of JCV infection but is not predictive of PML risk. Several informative and recent reviews of PML risk stratification are available (225, 466). Current risk information for natalizumab is available from Biogen-IDEC (https://medinfo.biogenidec.com).
The central feature of the pathology associated with PML is the infection of the oligodendrocyte with JCV, which leads to lysis of the cell. The infection spreads to surrounding oligodendrocytes and results in focal destruction of myelin. The infected oligodendrocytes have collections of viral particles in the nuclei, which give the appearance of inclusion bodies and loss of chromatin structure upon examination by light microscopy. The size of the nucleus may also increase by as much as 2- to 3-fold. Morphologically normal oligodendrocytes may also be infected with the virus, as seen by immunohistochemistry or in situ hybridization. These cells are usually present in areas where the myelin appears to be normal. The multifocal nature of the lesions suggests a hematogenous spread of the virus to the brain. To a lesser extent, astrocytes are also infected with JCV. These cells appear large, and the nuclei are irregular and lobulated and appear premitotic but do not become neoplastic. These cells have been described using the term “bizarre.” Reactive astrocytes are also present, which is a nonspecific finding. Although neurons themselves are rarely productively infected by JCV, demyelination leads to axonal dysfunction, and the demyelinated axon is susceptible to injury by cellular products released by the glial cells. Axonal injury can result in a retrograde loss of the neuronal cell body. Loss of neurons is likely permanent.
Invading macrophages are commonly seen in the centers of lesions. They act as scavengers and are often laden with myelin debris in the sites of lesions. Macrophages and microglia are not infected by JCV. In PML patients with HIV infection, there can be massive necrotic lesions with infiltration by HIV-infected macrophages (536).
Lymphocytes are not typically seen in PML lesions except if there is restoration of the immune system leading to an immune reconstitution inflammatory syndrome (IRIS). Under these circumstances, CD8+ T cells are the predominant cell type and are present in perivascular regions at sites distant from areas with JC virus infection as well as in focal collection in the parenchyma in proximity to JCV-infected cells (511, 544). The presence of cytotoxic T cells against JCV is considered to be a good prognostic sign. It is postulated that the rapid restoration of the immune system may lead to an expansion of activated T cells, which may contribute to the massive inflammatory response seen in patients with IRIS and can result in significant morbidity and mortality.
Symptoms of PML are usually insidious in onset. The initial symptoms often go unnoticed by patients and family and are brought to medical attention only when there is significant impairment of cognition or motor function. It is not unusual for the initial presentation to be mistaken for a stroke. However, symptoms continue to progress gradually over days to weeks. The differential diagnosis in part depends upon the underlying condition. In patients with HIV infection, the presence of focal symptoms such as hemiparesis, visual deficits such as loss of vision on one side, aphasia (i.e., inability to either comprehend or express speech), or ataxia help differentiate PML from HIV-associated neurocognitive disorders (HAND) or HIV encephalitis. In patients with multiple sclerosis who develop PML while on treatment with natalizumab, the differential diagnosis of PML from a relapse of multiple sclerosis can sometimes be challenging, since motor symptoms, ataxia, and visual abnormalities can occur with either disorder. However, a change in personality or cognitive abilities, new-onset seizures, or aphasia suggests the likelihood of PML and warrants further investigation. Since PML is multifocal in nature, the associated clinical manifestations may also represent lesions in more than one region of the brain. Although PML can affect any part of the brain, it has a predilection for the posterior regions of the brain, including the brain stem, cerebellum, and occipital lobe, and hence the clinical symptoms mirror these effects (35).
In recent years the realization that restoration of the immune system is the best mode of treatment for PML has led to the use of aggressive antiretroviral therapy in patients with HIV infection and rapid withdrawal of immunosuppressive therapy in patients with autoimmune diseases or organ transplants. This results in an influx of lymphocytes into the brains of these patients with PML, potentially leading to the clinical syndrome of immune reconstitution inflammatory syndrome (IRIS). This is a T cell-mediated encephalitis, and although the cells are needed to control JCV replication, paradoxically, the associated massive inflammatory response can result in injury to the surrounding brain tissue and result in deterioration of the neurological symptoms. Occasionally the inflammation can be so severe so as to lead to massive swelling of the brain, resulting in herniation and death of the patient (488).
Magnetic resonance imaging (MRI) of the brain can show characteristic features diagnostic of PML. They typically show multiple high-signal-intensity lesions on T2-weighted and FLAIR sequences (Fig. 4). They usually involve the uncinate fibers and have a predilection for the posterior parts of the brain. The lesions typically spare the gray matter, since JCV infects mainly oligodendrocytes and some astrocytes. On T1-weighted images, the lesions may appear hypointense. Since these lesions occur when the patients are immunosuppressed, no enhancement of the lesions is seen when the contrast agent gadolinium is administered. However, patients who develop IRIS have variable degrees of enhancement. Patients with multiple sclerosis typically develop prominent enhancement with IRIS, since the immune system is intact when the drug is withdrawn, while patients with HIV infection may show subtle enhancement, since the cell counts in the blood are low and hence fewer cells enter the brain. Magnetic resonance spectroscopy shows an increase in choline and lipids suggestive of gliosis and myelin breakdown; N-acetyl acetate is decreased, suggestive of axonal injury (72).
Although antibody testing in plasma can be used to assess the risk of PML prior to onset or use of immunomodulatory therapy, it is not currently of diagnostic significance after the onset of PML. Instead, the confirmatory test for suspected PML is the demonstration of the presence of JCV DNA in the cerebrospinal fluid (CSF) or brain. Detection of the virus DNA in the CSF by PCR is of diagnostic significance but detection in the blood is not, as viremia is sometimes detectable in the absence of PML, while a percentage of PML patients are not viremic (515). The specificity of quantitative PCR (qPCR) can be 100% by targeting unique sequences within the JCV T antigen that are necessary for infection. There is no cross-reaction to the human polyomavirus BK virus (BKV or BKPyV) (428) or in specimens from patients with CNS diseases other than PML (51). The detection sensitivity of some qPCR assays can be as low as 10 copies/ml (428). Along with clinical evidence and MRI, qPCR results help confirm a diagnostic result for PML. Detection of JCV DNA by qPCR and the copy number of viral DNA can be used for diagnosis in conjunction with clinical and radiographic findings. Patients with lower, or decreasing, levels of JCV genomes in the CSF after therapy may show better longevity (108, 109), but the prognostic significance of the viral load of JCV in the CSF has not yet been fully established (51).
In patients with untreated HIV infection, the viral copy numbers are usually quite high and easily detectable, but in patients with multiple sclerosis where the immune system is relatively intact, the copy numbers of the virus can be quite low and more difficult to detect (430). In the era of HAART, HIV patients with PML and low or undetectable JCV copy number have been described with increased frequency (319, 320), and HAART treatment is correlated with a reduction in the viral load of JCV in the CSF (108, 166). Brain biopsy is sometimes indicated in patients where the CSF cannot be obtained or is inconclusive and the MRI is not characteristic of the disease. In the brain tissue, the infected cells can be demonstrated by immunohistochemistry, in situ hybridization, or PCR analysis (8). Other changes in the CSF are nonspecific, with a mild increase in protein but a normal cell count and glucose. Interestingly, despite immune reconstitution, some patients may not clear the virus completely, and it may persist in the CSF (430). While the significance of this persistent virus is not clear, it indicates that JCV can remain in the brain for long periods despite reconstitution of the immune system or frank IRIS.
The prognosis of PML is generally poor. In the pre-AIDS era and before antiretroviral drugs were available, death was nearly universal, with an average survival of 9 months in non-HIV patients (474) and 2 to 4 months in patients with HIV infection (36). Although survival of patients with PML has improved due to the use of antiretroviral drugs, early recognition, and improvement in diagnostic techniques, the mortality rate is still nearly 50% in HIV-infected patients, and while in the multiple sclerosis population the mortality is lower, the morbidity is severe in the survivors.
Currently, there is no specific antiviral drug against JC virus. Anecdotal reports of response to various treatments are scattered throughout the literature. However, all controlled studies have failed to show any efficacy of the drugs tested against PML. This includes cidofovir (CDV), cytosine arabinoside (Ara-C), and mefloquine (55). The best treatment for PML is the restoration of the immune system, although even this is not ideal, since it can lead to IRIS. It is hence recommended that while interventions are being made to restore the immune function, such as initiation of antiretroviral agents in HIV-infected individuals and removal of the offending chemotherapeutic agent, the patients should be closely monitored for the development of IRIS and treated with steroids accordingly (219).
Many broad-spectrum nucleoside analog chemotherapeutics that target DNA replication have been used to inhibit JCV replication in PML patients without much success, including cytosine arabinoside (Ara-C), adenosine arabinoside (Ara-A), azidothymidine (AZT), acyclovir (ACV), and cidofovir (CDV) (315). Nucleoside analogs interrupt RNA and DNA synthesis and therefore can be used as potent antiviral agents; however, they are also highly toxic due to interruption of host RNA and DNA synthesis. Ara-C and, to a lesser extent, AZT have been shown to limit JCV replication in a tissue culture model (196) but to vary in their ability to inhibit JCV in PML patients in vivo. Ara-C monotherapy (7, 59, 109, 159, 274, 313, 364, 376, 393) or in combination with CDV (81, 184, 492, 520), methotrexate (162), or interferon (159, 189, 471) has been reported with positive prognosis in some cases and death in other cases (18, 89, 180, 195, 204, 346, 404, 462, 502). The ACTG 243 clinical trial showed no benefit in survival rates for PML patients treated with either intrathecal or intravenous administration of Ara-C. It was unlikely, however, that the drug ever reached the multiple plaque lesions in the AIDS/PML patients in that trial (180).
Cidofovir (CDV), an acyclic nucleotide phosphonate analog of deoxycytosine monophosphate effective for treatment of cytomegalovirus retinitis, has shown antiviral activity for nonhuman polyomaviruses using in vitro cell cultures (16), but most cases of CDV use for treatment of PML demonstrate no benefit (109, 312, 546). More recently a hexadecyloxypropyl lipid conjugate of CDV, commercially known as CMX001, was shown to inhibit JC virus replication in cell cultures derived from human fetal brain (174, 216). In 2010, CMX001 was used in combination with interleukin-7 treatment in a patient with PML and idiopathic CD4+ lymphocytopenia, resulting in a significant reduction of viral loads and improvement in clinical symptoms over 8 weeks (382). Importantly, both in vitro studies reported significant levels of toxicity as measured by decreasing cell viability with increasing concentrations of CMX001 treatment, and toxicity remains an important consideration for CMX001.
Another pathway targeted for treatment of PML includes inhibition of virus entry into the host cell and presumably of spread between cells, limiting the progressive nature of the disease. JCV enters into host cells through binding of the virus to the primary receptor α2,6-linked sialic acid moieties (125, 361) and the secondary receptor serotonin receptor 2A (5HT2AR) (130, 361) on the cell surface. Because sialic acid moieties are common among all cell types, including cells not permissive to JCV infection, blocking JCV binding to this molecule is likely not an effective target for therapeutic intervention. More recent studies have focused on blocking JCV binding to 5HT2AR on permissive cells in the brain using serotonin receptor agonists (385). Blocking access to 5HT2AR using antibodies or the serotonin receptor agonists chlorpromazine and clozapine was effective in limiting JCV infection in human brain-derived cell cultures (21, 30, 130, 371, 373, 442); however, these drugs have serious side effects and toxicity issues. Newer antipsychotics, including zisprasidone, risperidone, and olanzapine, were shown to inhibit JCV infection up to 10-fold more potently than the previously studied agonists in an in vitro system (11). Based on these in vitro results, serotonin receptor agonists have been used to treat PML with various degrees of success. Treatment of PML with mirtazapine alone (263) or in combination with Ara-C (520) was associated with a favorable outcome in some patients, while combined chlorpromazine and CDV therapy was ineffective in lowering JCV levels in either the CSF or plasma (392). Subsequent studies using human brain microvascular endothelial cells (74) and human fetal progenitor-derived astrocytes and oligodendrocytes (343) showed that JCV infection of these cells occurs independent of 5HT2AR expression, suggesting that 5HT2AR is not sufficient or essential for JCV infection of certain cell subsets in the human brain. Further studies are warranted to determine the efficacy of serotonin receptor agonists as treatment for PML.
The antimalaria drug mefloquine was shown to inhibit JCV replication using in vitro cell culture models derived from human fetal brain in an attempt to identify FDA-approved, commercially available drugs/biologically active molecules with antiviral activities against JCV (56). Mefloquine is known to cross the BBB and accumulate in the brain, where JCV infection is pathological (221, 388), but has been associated with neurotoxicity (493). Several independent case reports showed that mefloquine treatment of PML was successful in reducing the viral burden in the brain and was associated with improvement of clinical symptoms (168, 244). However, as reported at the 2011 annual meeting of the American Academy of Neurology, a multicenter clinical trial supported by Biogen-IDEC Inc. and Elan Pharmaceuticals failed to show a reproducible reduction in the JCV DNA in PML patient CSF or reduced clinical progression of PML in response to mefloquine treatment (150).
Although development of PML no longer guarantees fatality, as described above, the prognosis is poor and therapeutic options are few and regularly ineffective. Withdrawal from immunosuppressants in non-AIDS PML and HAART therapy in AIDS-related PML have been associated with immune reconstitution in the brain and control of viral replication. Rapid immune reconstitution is important to CNS immunosurveillance and control of JCV replication (33, 129, 161, 223, 249); however, IRIS itself can be a serious, often fatal outcome. Low T cell counts and high numbers of copies of JCV DNA in the CSF at the time of PML diagnosis are clear risk factors for death (66, 128, 160, 249). Early use of five-drug combined antiretroviral therapy after PML prognosis has been shown to improve survival, which is associated with recovery of anti-JCV T cell responses and reduction of JCV DNA in the CSF (160). Adoptive transfer of JCV antigen-specific cytotoxic T cells, generated after in vitro stimulation with the viral T and VP1 proteins, concurrent with citalopram (a serotonin reuptake inhibitor) and CDV treatment resulted in clearance of JCV from the CSF and improvement of clinical symptoms in a hematopoietic cell transplant recipient with PML (27). In addition, bioenergetic parameters that reflect T cell immunocompetence, such as intracellular CD4+-ATP concentration, has been shown to inversely correlate with risk of infections during immunosuppressive therapy, including infections with the human polyomaviruses BKV (28) and JCV (178). These studies continue to demonstrate the essential role that the immune system plays in both development of PML and control of JCV replication. As a result, the development of future therapeutics for PML should focus not only on blocking viral replication but also on reconstituting an effective T cell response against the virus in the brain.
While the typical characteristics of PML are those of a white matter disease, productive infection of granule cell neurons by JCV has been reported. Changes in the cerebellar granule layer during PML have been observed for over 50 years (419), and these include enlarged and hyperchromatic nuclei (420), which are found in approximately 5% of PML patients. For many years it was unclear whether these cells were directly infected with JCV or were damaged collaterally as a result of the destruction of glial cells by JCV.
In 2003, productive infection by JCV of granule cell neurons in the cerebellum in an HIV-positive patient with PML was described (127). This individual presented with pyramidal tract and cerebellar dysfunction. Autopsy revealed classic PML in the frontal lobe and productively infected neurons expressing viral proteins in the internal granule cell layer, as demonstrated by immunohistochemistry and electron microscopy. Infected neurons which did not express viral proteins were also detected by in situ hybridization.
Subsequently, JCV was found in the brain of a patient with cerebellar atrophy but no detectable white matter PML lesions (251). This pathology, termed JC virus granule cell neuronopathy (JCV-GCN), was proposed to be a novel clinical syndrome distinct from PML. JCV-GCN has been since described in both HIV-positive and HIV-negative patients (175, 188, 228, 448, 487). In the original case study, it was determined that the NCCR was type IS, with no “b” and “d” sequence blocks as found in archetype and no repeat structure but with some other small insertions that did not correspond to sequences found in any known JCV isolate. A more recent case involving an AIDS patient with cerebellar atrophy also showed rearrangement of the NCCR, also with a type IS/IIS arrangement (with no sequence block “b” and only a partial piece of block “d”). Comparison of CSF-isolated virus and cerebellar virus NCCRs showed differences and changes in transcription factor binding sites (426). Thus, it appears that alternative transcription factor binding sites may play a role in JCV-GCN, further demonstrating the importance of the NCCR sequence in JCV cell type specificity.
Alternatively, entry into neurons or postentry steps such as transport, uncoating, or assembly may be changed by differences in JCV capsid proteins. In multiple JCV-GCN case reports, deletions and frameshift mutations in the C terminus of the JCV VP1 gene have been found (93, 94) when matched to the Mad-1 variant as well as to a hemispheric white matter isolate from the index case of JCV-GCN.
JCV-GCN is a productive infection of granule cell neurons and should be considered in cases of cerebellar atrophy. Symptoms are indicative of subacute or chronic cerebellar dysfunction, including gait abnormalities, dysarthria, and incoordination (486). MRI findings of cerebellar atrophy in the absence of white matter lesions and positive PCR of the CSF indicate the possibility of JCV-GCN (486). Diagnosis is complicated by the possibility that both JCV-GCN and classic PML pathologies can occur simultaneously. The finding of JCV-GCN in conjunction with classic PML requires cerebellar biopsy showing a lytic infection of granule cell neurons. It is possible that JCV-GCN, or productive infection of neurons in general, is a frequent complication in classic PML cases, with a report of 79% of PML cases showing infection of granule cell neurons (542). It is also possible that classic PML, with lesions of undetectable size, may be present in the reported cases of JCV-GCN.
JCV protein expression in cortical neurons neighboring white matter lesions of PML patient samples has also been detected, although the predominant expression was of T antigen, indicating the possibility of an abortive or restrictive infection in these cells (545). In one case, termed JCV encephalopathy, lytic infection by JCV of cortical pyramidal neurons with limited demyelination was described (543). Another rare pathology of JCV, JCV meningitis, has also been reported (40, 512). While these case reports are of interest in that they demonstrate a possible expansion of permissive cell types and clinical pathology of JCV, this review focuses on classical PML and PML-IRIS, as this has been the focus of the majority of research on JCV. More detailed information on rare JCV-associated diseases can be found in a recent review article (486).
In addition to its role in the development of PML, several studies have underscored the ability of JCV to transform cells in culture and induce tumors of neural origin in several experimental animal models, such as golden Syrian hamsters and rats (114, 287, 288, 374, 523).
Early studies also confirmed that JCV caused tumors in nonhuman primates, specifically New World monkeys, including owl and squirrel monkeys. While several different kinds of tumors form in JCV-inoculated rodents, JCV-induced tumors in owl and squirrel monkeys are almost exclusively gliomas. These gliomas develop 14 to 36 months after intracranial inoculation with JCV, with a long period of asymptomatic “latent” infection followed by rapid tumor growth and progression, neurological impairment, and death (200, 287, 288, 329). New World monkeys have been used to study the immune response to JCV-induced tumors (524), as well as to develop radiographic diagnostic procedures for human astrocytomas (199).
Most JCV-induced brain tumors in monkeys have integrated viral DNA (335), often in a head-to-tail configuration of at least two copies (336), and express the large T antigen (298). T antigen did not bind the p53 tumor suppressor protein (304). Interestingly, in another case, JCV T antigen was found to bind p53, indicating that there were structural differences between T antigen in this tumor and that from the previous study (308). This was also the first case of p53 described in a primate brain tumor. In this study, a glioma was induced in an owl monkey inoculated with JCV, and recipient owl monkeys were inoculated intracranially with a tumor cell suspension of the explanted glioma tissue. One monkey (termed owl monkey 586), developed an astrocytoma, which, upon explant, was shown to have both integrated and free viral DNA with an NCCR corresponding to the Mad-4 variant. These cells also produced JCV T antigen which bound p53 and produced infectious virus particles. Studies of gliomas in New World monkeys demonstrated the possibility that JCV could cause gliomas in nonhuman primates and that, at least in rare cases, these transformed cells could produce infectious virus. These findings provided an impetus to determine whether an association between JCV and human cancer could be found.
Recent studies point to the association of JCV with human cancer, including brain tumors, colon cancer, and others, but it is important to note that there is a substantive variation between different laboratories as to the frequency of association of cancer and the presence of JCV (170, 423). It is now evident that the oncogenic activity of JCV is closely linked with the expression of the portion of its genome carrying the early gene, whose product, large T antigen, has the capacity to associate with several important cellular proteins that are essential for the control of cell growth and proliferation (117, 534). Direct evidence for tumorigenicity of T antigen comes from studies of transgenic animals encompassing the JCV early genome, demonstrating that in the absence of viral replication, expression of T antigen induces a broad range of tumors of neural origin, including tumors of glial origin, primitive neuroectodermal tumors (PNETs), abdominal neuroblastoma, malignant peripheral nerve sheath tumors (MPNSTs), and pituitary tumors (114, 145, 256, 257, 456, 460). Histological and molecular studies of tumors from these transgenic mice revealed the interaction of T antigen with well-characterized tumor suppressor proteins, including p53 and members of the pRb family, in these tumors and dysregulation of related cell cycle controllers, including cyclins, cdks, and p21WAF (256). Furthermore, the association of T antigen with other tumor suppressor proteins, including the neurofibromatosis 2 gene product (merlin or NF2), has been observed in JCV transgenic mice with MPNSTs (456).
Extensive study of tumors derived from T antigen transgenic mice have elucidated several other pathways that may contribute to the development of PNETs caused by JCV. In one series of studies, it was demonstrated that T antigen, by association with β-catenin, results in the stabilization of β-catenin and its nuclear localization in the tumor cells. Interestingly, upregulation of the expression of several targets for β-catenin was observed in these cells, including cyclin D and c-myc (158). On the other hand, in the presence of T antigen, the cross-communication of β-catenin with G protein kinases, including Rac, in the cell membrane was found to activate alternative oncogenic signal transduction pathways involving JNK, Ras, NF-κB, and others (39).
In addition to β-catenin, JCV T antigen also interacts with insulin receptor substrate-1 (IRS-1), a key mediator of the insulin-like growth factor-1 (IGF-1) pathway. This interaction promotes nuclear localization of IRS-1 in mouse medulloblastoma cell lines that are positive for T antigen (265). While the exact biological importance of this event remains to be determined, recent studies suggest a role for nuclear IRS in chromosomal instability (414, 496). Thus, current evidence indicates that JCV T antigen, by altering the biological activities of several tumor suppressor proteins, as well as proteins involved in multiple signaling pathways, can initiate a cascade of events that leads to uncontrolled cell proliferation in experimental animals containing the JCV genome and in tumor cells derived from these animals.
Several reports indicate that different forms of human cancer, most notably medulloblastoma, are positive for the presence of the JCV genome and the expression of JCV T antigen as well as the late auxiliary protein agnoprotein, but not viral capsid proteins, in the tumor tissue (113, 255). These observations remain somewhat controversial, as other reports indicate that JCV is not the causative agent of these tumors (186, 241, 349, 509). Interestingly, p53 was reported to be colocalized in the nucleus with T antigen, suggesting that the interaction of T antigen with p53 may abrogate the role of p53 in control of cell proliferation. In addition, IRS-1 was found in the nuclei of these tumor cells, as was also the case for tumors arising in mice that are transgenic for the JCV early region (115). The detection of JCV agnoprotein, which exhibits the ability to dysregulate the cell cycle via p53 and cyclin B and to impair DNA repair through Ku70/80 (100, 101), has also been reported in human tumor cells (113). Agnoprotein was reported to be localized in the perinuclear region, as is also observed in PML (236).
The JCV genome has also been reported for several tumors of nonneural origin, including colorectal cancer, gastric cancer, esophageal carcinoma, and lymphoma (65, 69, 111, 116, 131, 132, 157, 193, 261, 280, 282, 340, 350, 365, 417, 447, 449, 450, 513, 549), although it should again be noted that there are also reports to the contrary (163, 291, 363). Consistent with observations in tumors of neural origin, examination of these tumor cells showed expression of T antigen, and in some cases agnoprotein, in tumor cells, where it was colocalized with p53 and β-catenin (65, 111, 116, 131, 261). While the importance of these observations in regard to the initiation of transformation or in maintenance of tumors remains to be established, the detection of JCV viral proteins in these tumor cells opens the possibility that JCV T antigen and possibly agnoprotein may accelerate the development and progression of tumors by inactivating tumor suppressors and/or dysregulating signaling pathways.
The evidence outlined above concerning the association of JCV with human cancer and the parallel molecular findings observed in JCV-induced tumors in experimental animal models support a potential role for JCV in the pathogenesis of human tumors harboring JCV DNA sequences and expressing JCV oncoproteins. Nevertheless, consensus concerning the extent of involvement of JCV with human tumors is still lacking, and this is presumably due to differences in technical issues that exist between different laboratories, which have recently been reviewed in detail (117). It is possible that the expression of T antigen in JCV-positive cells may not persist during the course of tumor cell maturation and progression. In fact, several studies have shown that T antigen-positive cells may lose expression of T antigen after extensive passage in culture. The loss of T antigen expression appears to have no significant impact on the doubling time of the cells. Interestingly, the gene sequences for T antigen remain intact in these cells, suggesting that loss of T antigen may result from other mutations in the JCV genome, modification of the viral DNA and/or RNA, or rapid turnover of T antigen due to its degradation. These observations have led to the speculation that transient expression of T antigen reprograms pathways that control cell growth and proliferation by dysregulating factors involved in the control of the cell cycle and chromosomal stability such as p53. Examination of tumors from JCV transgenic mice indicates that expression of T antigen is observed in only a fraction of tumor cell nuclei, suggesting extinction of T antigen expression in a subpopulation of tumor cells due to genomic instability or acquired mutations that promote uncontrolled cell proliferation (535). Similarly, it is conceivable that T antigen may also be downregulated in human tumor progression. While possible early events in the genesis of tumors containing the JCV DNA sequence have yet be elucidated, it may be envisioned that JCV involvement in tumorigenesis may be caused by abortive infection, a rare event, which then leads to clades of cells that, upon T antigen and perhaps agnoprotein expression, enter the cell cycle, progressively grow, and become transformed. In this respect, T antigen may function by targeting tumor suppressors such as p53, pRb and NF2, and signaling factors, including IRS-1 and β-catenin, as determined in the molecular experiments with JCV early transgenic mice and human tumors described above. In this way, JCV may initiate the formation of, and perhaps contribute to the progression of, tumors of neural origin. Thus, further investigation of the possible role of JCV in human brain tumors is warranted, and to fully investigate a role for JCV with neural and nonneural tumors, large-scale epidemiological studies are necessary.
In the years since the advent of the AIDS pandemic, PML has become a growing concern, and the incidence of this disease even after the widespread use of HAART remains high. With the increasing use of biological immunomodulatory therapies for autoimmune and other inflammatory conditions, the incidence of PML has increased dramatically. Currently, the PML incidence in HIV-infected patients is approximately 3%. As of 1 February 2012 there were 207 reported cases of natalizumab-related PML, with an overall incidence estimated at approximately 1 in 500, although the incidence of PML is roughly 1 patient in 250 after 24 months of natalizumab treatment and 1 in 100 after 24 months of treatment in addition to prior immunosuppressant use (current statistics for natalizumab-related PML are available at https://medinfo.biogenidec.com). A registry is being established through the Laboratory of Molecular Medicine and Neuroscience, National Institute of Neurological Disorders and Stroke, to increase monitoring and information on the incidence of PML in patients with various underlying diseases.
JCV maintains a restricted host range. Within the susceptible host, JCV can replicate in a small subset of cell types, which makes the continued study of JCV both extremely important and very challenging. Only cells of human origin are susceptible to productive lytic infection by JCV. This has precluded the development of an appropriate animal model for PML, and studies of pathogenicity are dependent on development of cell culture models for JCV.
Viral binding to host receptors may partially account for cell type specificity and disease progression, especially in light of evidence that some changes in VP1 structure are associated with progression to PML or other rare JCV-induced diseases. JCV is widespread globally, but distinct populations show differences in infection incidence as well as predominant JCV subtypes. VP1 changes can divide JCV into 13 distinct types, with other subtypes distinguished by changes in the other coding regions. These subtypes have distinct ethnic and geographic distributions and have been used for various purposes, from mapping ancient population movements to identifying cadavers. Subtypes also show differences in PML risk, as VP1 type 2B is associated with a slightly higher PML risk, whereas type 4 shows a slightly lower PML incidence. More importantly, changes in VP1 that are not associated with distinct subtypes are associated with virus from PML patients but rarely with that from healthy people. Known PML-associated changes in VP1 cluster in the receptor binding region and likely influence infection to some degree.
Although VP1 changes are potentially associated with PML, much remains to be determined about how these changes influence PML risk. It is clear that variations in the NCCR sequence leading to unique and/or duplicated transcription factor binding sites greatly enhance the risk of progression to PML. Host and cell type specificities are controlled primarily at the level of transcription and replication factors. JCV is able to bind and enter most cell types studied, but in nonsusceptible cell types, the viral life cycle is blocked at the level of early gene transcription.
Because of differences in host factor availability, the host range of JCV is complex. In humans, JCV causes a slow, lytic infection, while in rodents and nonhuman primates, JCV infection leads to tumor formation, which is primarily due to the absence of DNA replication due to differences in host factors in these animal models. This leads to an overexpression of T antigen, where its potential to interfere with cell cycle progression is realized as malignant transformation.
It is unclear whether human cancer can also be caused by JCV infection, as there is conflicting evidence on both correlation and causation of malignant transformation by JCV. The most likely candidates are glioblastomas, colorectal cancer, and lymphomas, as these have the most evidence in support, but it is clear that continued study is needed in order to determine what, if any, role JCV infection plays in human cancer. In addition to the study of direct roles of JCV in human cancer, the virus is also a useful tool to dissect mechanisms of transformation as well as host antiviral defense and the intersection of these mechanisms with tumor suppression.
JCV-induced oncogenesis is an important area of study; however, the primary disease caused by JCV is PML. As there is no animal model for PML, the continued advance of knowledge about JCV transcription and replication is dependent on advances in cell culture techniques able to support JCV replication, as well as tissue from PML patients and healthy donors. An interesting avenue of investigation is the differences in immune function that lead to PML. CD4+ T cells are important, leading to an antiviral CD8+ T cell response. It is unclear whether humoral immunity plays a role in controlling infection, as antiviral antibody titer does not correlate with PML risk.
Although over half of the global population is infected with JCV, only a small subset of immunosuppressed individuals develop PML, and even in the most severe cases, such as in AIDS patients, the incidence of PML is less than 5%. This suggests that not all underlying causes of immunosuppression are the same. AIDS patients vary both in their level of immunosuppression and in specific immune functions, such as immune signaling, activation, cellular migration, and surveillance. Elucidation of the differences in immune states between those immunosuppressed patients who develop PML and those who do not will lead to greater understanding of JCV and the immune system, as well as to therapies of increased effectiveness. These mechanisms will likely have to be determined through human studies, as recapitulation of immune responses in cell culture does not truly mimic in vivo responses, and there are no animal models suitable for immunological study of JCV and PML.
The discovery that JCV can infect cells of the immune system provided insight into the route and mechanism of infection and the migration of JCV to the brain. It has been proposed that JCV infects either circulating B cells or stromal cells in the tonsil, by inhalation or ingestion, early in life. Virus may then be disseminated to the bone marrow, kidneys, and/or brain, where JCV can remain latent for the life of the host. JCV has been found to infect CD34+ hematopoietic progenitor cells and primary B cells (341). It is probable that CD34+ hematopoietic progenitor cells function as a viral reservoir. Many viruses infect cells of the immune system, including human cytomegalovirus (hCMV). Interestingly hCMV also uses CD34+ cells as a site of viral latency (295, 332, 458, 519) and can transactivate JCV (538). Both viruses can be modulated by HIV (84, 270, 481), which has also been proposed to infect and establish latent reservoirs in CD34+ cells (68). All three viruses are also activated by factors involved in B cell differentiation or by the differentiation process itself (68, 270, 314). The interplay of JCV and other viruses that infect the immune system and brain is an area that requires further study, especially in that coinfection of certain viruses may be a risk factor for the development of PML.
Additionally, results supporting infection of CD34+ hematopoietic progenitors gives rise to the possibility that mobilization of these cells and differentiation in the periphery are a risk factor for PML. Natalizumab treatment for MS, which increases risk of PML, mobilizes CD34+ progenitors from the bone marrow to the periphery (49, 217, 558), and these mobilized cells show greater migration toward chemokine stimuli (217). HIV infection, as well as MS, for which natalizumab is a treatment, also lead to CNS damage and inflammation and thus to increased chemokine stimuli (194, 243, 530, 547), which may lead to greater migration of virally infected cells to the brain. Thus, it appears that changes in immune trafficking and maturation may play a large role in the development of PML.
These immune changes may also influence the rearrangements in the JCV NCCR required for development of PML. Archetype-like virus has been found rarely in PML brain, although it is almost the exclusive type of JCV excreted in the urine. The prevailing hypothesis is that the archetype NCCR is the form transmitted from person to person. In this model, the virus spreads through the body and remains latent in the kidney and bone marrow. Following deletions and rearrangements of NCCR sequence, which change, delete, and duplicate host transcription factor binding sites, JCV becomes fit to replicate in glial cells and thus cause PML. It is unclear in which compartment these changes occur, although both archetype and prototype-like sequences have been found in the bone marrow, making it a likely site of viral DNA rearrangements. Additionally, maturation of CD34+ cells to mature B cells leads to an upregulation in DNA recombination factors required for V(D)J recombination, making these cells an attractive site for the possibility of viral DNA recombination. Mobilization to the periphery and subsequent maturation of JCV-infected CD34+ cells could lead to these rearrangements, and thus an increased risk of PML, in immunosuppressed patients and those on immunomodulatory therapies such as natalizumab.
It is important to note that JCV may be found in several sites of latency. Infection of cells of the brain may occur at different times throughout the life of the host. PML may be caused by reactivation of latent virus in the brain due to decreased immune surveillance, as well as trafficking of new virus to the brain due to increased lymphocyte mobilization. Because of these factors, as well as increased viral replication, there may also be a number of variants of the NCCR found in the brain during PML. Next-generation sequencing and additional research may lead to further elucidation of the pathways of viral dissemination.
If the model in which JCV is latent in the bone marrow and mobilizes to the CNS upon changes to the immune system is correct, then progression from latent JCV infection to active PML likely requires four changes: (i) a decrease or change in immune function, (ii) rearrangement of the regulatory region to increase replication fitness, (iii) mobilization of infected cells from sites of latency to the brain, and (iv) upregulation of factors that increase JCV replication. This is illustrated in the case of Spi-B. Natalizumab decreases immune surveillance to the brain and, additionally, causes upregulation of Spi-B, a transcription factor involved in B cell development which is also expressed by glial cells, in the peripheral blood (281). TATA box-associated Spi-B binding sites in prototype-like viral variants are important for viral transcription (314). Thus, the combination of underlying disease, immune modulation and changes in trafficking, and upregulation of factors that result in increased viral transcription, in addition to changes in viral NCCR sequence, lead to an increased risk of PML. Investigation of a number of other cell type-specific factors, including NFI-X, that have been implicated in viral pathogenesis is required to come to a greater understanding of the risk factors for PML.
These risk factors are an increasingly important area of study in the viral pathogenesis of JCV. These factors may eventually be used as markers for PML risk and disease progression and may inform treatment of PML, as well as underlying conditions with a risk of PML. Stratifying patient risk is an important consideration being investigated in the case of immunomodulatory therapies, with the goal of finding alternative treatments for those at highest risk of developing PML while maintaining the use of these promising new therapies for those at lower risk of development of PML. Host factors may also be potential targets for new PML therapies.
Currently, there is no effective therapy for PML. In the case of AIDS, immune reconstitution is the only available attempt at “treatment,” and this course is often not available after severe immunosuppression. Even when immune reconstitution is possible, IRIS is a major risk and can lead to permanent damage or death. When PML occurs during immune therapies such as natalizumab treatment, plasmapheresis is the only course of action. This may also lead to IRIS, as well as increased symptoms of the underlying disease. Several investigational drugs for PML exist, including CMX001. CMX001 has been shown to inhibit JCV replication in cell culture (216), but it has toxicity and is approved only under review by the FDA for cause. Therapies with greater efficacy and specificity and with lower toxicity are urgently needed for the treatment of PML. Because of the steadily rising incidence and significant morbidity and mortality of PML, further investigation of the epidemiology, pathogenesis, and molecular biology of JCV, as well as potential therapies for PML, is of increasing importance.
W.J.A. is supported by NIH grants P01NS065719, R01NS043097, R01CA071878, and F32NS070687 (CN). K.K. is supported in part by NIH grants R01NS35000 and R01MH086358. M.W.F. is supported in part by the Intramural AIDS Research Fellowship from the NIH Office of AIDS Research. The Laboratory of Molecular Medicine and Neuroscience, as well as the Section of Infections of the Nervous System, is supported by the Division of Intramural Research of the NINDS.
We thank Wendy Virgadamo at Brown University for help in the preparation of Fig. 2 and Patrick Lane of ScEYEnce Studios for art enhancement. We also thank members of the LMMN for input and critical reading of the manuscript.
Michael W. Ferenczy is a postdoctoral research fellow in the Laboratory of Molecular Medicine and Neuroscience at the National Institute of Neurological Disorders and Stroke at the National Institutes of Health (NIH). He obtained his Ph.D. in molecular virology and microbiology in the laboratory of Dr. Neal DeLuca at the University of Pittsburgh School of Medicine, where he studied the role of the transactivating protein ICP0 in epigenetic control of herpes simplex virus 1 transcription. His current research interests involve the role of the NFI family of proteins in the JCV life cycle, as well as JCV-HIV interactions that lead to PML.
Leslie J. Marshall is a postdoctoral research fellow in the Laboratory of Molecular Medicine and Neuroscience at the National Institute of Neurological Disorders and Stroke at the National Institutes of Health (NIH). She obtained her Ph.D. in microbiology and immunology in the laboratory of Dr. David Ornelles at Wake Forest University, where she studied the molecular regulation of adenovirus latency in T lymphocytes involving the leukemia-associated protein RUNX1. Her current work as a research fellow in the laboratory of Eugene Major at NIH led to the novel observation that the B cell transcription factor Spi-B is expressed in human glial cells and is involved in the molecular regulation of JC virus gene expression in these cells. Her studies continue to support an important role for JC virus-infected lymphoid cells in the molecular pathogenesis of progressive multifocal leukoencephalopathy.
Christian D. Nelson is a postdoctoral research fellow in the laboratory of Professor Walter Atwood in the department of Microbiology, Cell Biology, and Biochemistry at Brown University. He obtained his Ph.D. in comparative biomedical sciences in the laboratory of Professor Colin Parrish at Cornell University, where he studied antibody neutralization and the in vitro stability and uncoating of canine parvovirus. His current research interests involve mapping the infectious entry pathway of JC polyomavirus and developing novel antiviral compounds that inhibit JC polyomavirus replication.
Walter J. Atwood is Professor and Vice Chair of the Department of Molecular Biology, Cell Biology, and Biochemistry at Brown University. His laboratory focuses on invasion of host cells by human polyomaviruses. His early work as a graduate student with Len Norking at the University of Massachusetts at Amherst led to the identification of the cellular receptor for SV40. As a postdoctoral fellow with Eugene Major at the National Institutes of Health, Dr. Atwood began work on the human polyomavirus JCV, focusing on transcriptional regulation of the virus in glial cells and B cells. Since joining the faculty of Brown University in 1995, Dr. Atwood has focused on understanding how human polyomaviruses engage host cell receptors to establish infection. His work has led to several major discoveries, including the characterization of the sialic acid-dependent infectious mechanisms for both JCV and BKV, characterization of the modes of virus entry into cells for both JCV and BKV, and identification of a JCV receptor complex.
Avindra Nath received his M.D. degree from Christian Medical College in India in 1981 and completed a residency in neurology at the University of Texas Health Science Center in Houston, followed by a fellowship in multiple sclerosis and neurovirology at the same institution and then a fellowship in neuro-AIDS at NINDS. He held faculty positions at the University of Manitoba (1990 to 1997) and the University of Kentucky (1997 to 2002). In 2002, he joined Johns Hopkins University as Professor of Neurology and Director of the Division of Neuroimmunology and Neurological Infections. He joined NIH in 2011 as the Clinical Director of NINDS, the Director of the Translational Neuroscience Center, and Chief of the Section of Infections of the Nervous System. His research focuses on understanding the pathophysiology of retroviral infections of the nervous system and the development of new diagnostic and therapeutic approaches for these diseases.
Kamel Khalili is a Professor in the Department of Biology, as well as the Department of Microbiology, at Temple University and holds an adjunct professorship at the University of Milan in Milan, Italy. He received his Ph.D. in microbiology from the University of Pennsylvania School of Medicine, where he studied regulation of human actin genes in adenovirus-infected cells. Following postdoctoral work at the Wistar Institute and the National Cancer Institute, he joined the faculty at the Jefferson Medical College of Thomas Jefferson University. Later, he became the Founder and Director of the Center for Neurovirology and Neurooncology at MCP Hahnemann University School of Medicine in Philadelphia, PA. Dr. Khalili is a Professor and the Founding Chair of the Department of Neuroscience and the Director and Founder of the Center for Neurovirology at the Temple University School of Medicine in Philadelphia, PA. His research program focuses on the viral oncology and the molecular biology of neurotropic viruses, with the goal of understanding the molecular basis of virus-induced or associated neurodegenerative diseases and cancers of the central nervous system.
Eugene O. Major received the Ph.D. degree from the University of Illinois Medical Center in infectious diseases and microbiology and was a part of a team that established a research center there on the genetics of viruses that cause cancer. Following academic appointments as Associate Professor at the University of Illinois Medical School and the Loyola University Medical School in Chicago, where he was also Associate Dean of Graduate Programs, Dr. Major joined the National Institute of Neurological Disorders and Stroke (NINDS) at the National Institutes of Health (NIH) in 1981. He has developed a world-recognized translational research laboratory in the Intramural Program focusing on mechanisms of viral pathogenesis in the human nervous system, which includes JC virus-induced demyelination, progressive multifocal leukoencephalopathy, and HIV-1/AIDS-associated encephalopathy. Dr. Major's research program has also established a novel human brain-derived progenitor cell population whose lineage can be directed to all major cell types of the brain, similar to embryonic stem cells.