Hans H. Hirsch, Department Biomedicine - Haus Petersplatz, University of Basel, Petersplatz 10, CH-4003 Basel, Switzerland. e-mail: email@example.com
JC polyomavirus (JCPyV) was the first of now 12 PyVs detected in humans, when in 1964, PyV particles were revealed by electron microscopy in progressive multifocal leukoencephalopathy (PML) tissues. JCPyV infection is common in 35–70% of the general population, and the virus thereafter persists in the renourinary tract. One third of healthy adults asymptomatically shed JCPyV at approximately 50 000 copies/mL urine. PML is rare having an incidence of <0.3 per 100 000 person years in the general population. This increased to 2.4 per 1000 person years in HIV-AIDS patients without combination antiretroviral therapy (cART). Recently, PML emerged in multiple sclerosis patients treated with natalizumab to 2.13 cases per 1000 patients. Natalizumab blocks α4-integrin-dependent lymphocyte homing to the brain suggesting that not the overall cellular immunodeficiency but local failure of brain immune surveillance is a pivotal factor for PML. Recovering JCPyV-specific immune control, e.g., by starting cART or discontinuing natalizumab, significantly improves PML survival, but is challenged by the immune reconstitution inflammatory syndrome. Important steps of PML pathogenesis are undefined, and antiviral therapies are lacking. New clues might come from molecular and functional profiling of JCPyV and PML pathology and comparison with other replicative pathologies such as granule cell neuronopathy and (meningo-)encephalitis, and non-replicative JCPyV pathology possibly contributing to some malignancies. Given the increasing number of immunologically vulnerable patients, a critical reappraisal of JCPyV infection, replication and disease seems warranted.
The JC polyomavirus (JCPyV) was discovered in 1964 as the first of 12 polyomavirus (PyV) species identified in humans thus far . However, only 4 of the 12 PyVs have been consistently linked to human pathologies: BKPyV- to PyV-associated nephropathy (PyVAN) and PyV-associated hemorrhagic cystitis (PyVHC); MCPyV to Merkel cell carcinoma and TSPyV to trichodysplasia spinulosa of the skin; and JCPyV to progressive multifocal leukoencephalopathy (PML) and rarely PyVAN [2, 3]. In fact, JCPyV has been discovered in the clinical pathology of PML, when Gabriele Zu Rhein and Sam Chou detected intranuclear PyV particles by electron microscopy of ultrathin sections of affected brain tissues (Fig. 1). The PyV particles formed crystalloid arrays, but were also scattered and in filamentous order . Interestingly, the patient was a 67-year-old woman who had suffered from chronic asthma for the last 30 years of her life. Subsequently, the virus was isolated by cell culture from post-mortem PML tissue of a patient bearing the initials J. C. permitting further virological, molecular, and epidemiological characterization [4-6]. At that time, no information on PyVs in humans was available, but some animal PyVs were known: The murine PyV (MPyV) discovered in 1953 in newborn mice with multiple tumors that provided the name (poly-, greek multiple; -oma, tumors); the simian vacuolating virus SV40 discovered in 1960 as a potentially oncogenic contaminant of polio- and adenovirus vaccines with millions of human exposures ; and the baboon polyomavirus SA12 in 1963. Therefore, the discovery of JCPyV raised important questions, which were then as critical as they are now today for the newly detected PyVs in humans: Is this a zoonotic infection with an animal reservoir or a human PyV? How is this virus related to other PyVs and what are the specific molecular genetic characteristics? What is the frequency of infection and how is it transmitted? Does it cause disease(s), and what are potential needs and measures for prevention and treatment?
It is clear today that JCPyV is highly specific to humans with only one major serotype, but at least seven major genotypes. The route of transmission is still unresolved, but JCPyV infects approximately half of the general population. JCPyV persists in the renourinary tract and viral replication is detected as asymptomatic viruria in about one third of the infected immunocompetent individuals. The key JCPyV diseases occur in the context of immunological failure to adequately control viral infection in parenchymal sites of the body, namely, the central nervous system (CNS) and the kidney, but possibly also other sites including the colon. After the wave of PML in HIV-AIDS patients in the era before combination antiretroviral therapy (cART), a resurgence of PML has occurred after widespread use of different biological and immunosuppressive therapies. Moreover, cases of JCPyV-mediated PyVAN in kidney transplants have appeared as well as other JCPyV-associated pathologies including possibly some forms of cancer. Thus, JCPyV remains challenging and a review of JCPyV biology and the clinical implications of JCPyV infection, replication, and disease seem warranted.
Virion and genome organization of JCPyV
JCPyV virions are non-enveloped particles of approximately 42-nm diameter of T = 7 icosahedral symmetry and hence, morphologically indistinguishable from other PyV particles. The capsid consists of 360 VP1 molecules spontaneously assembled from 72 VP1 pentamers and mediates host cell receptor binding for uptake. Inside the VP1 shell, the minor capsid proteins, the myristoylated VP2 and the VP3 are found in a ratio of 1:1:5 relative to VP1 and probably play a role not only in packaging, but also in uncoating, and delivery of the JCPyV genome to the host cell nucleus [7-9].
The JCPyV genome is a circular double-stranded DNA of ~5100 base pairs wrapped around histones in a nucleosomal core-like organization. Akin to all PyVs, the overall structure of the JCPyV genome can be divided into three major regions (Fig. 2):
The non-coding control region (NCCR) of ∼400 bp harboring the origin of viral DNA replication ori, TATA- and TATA-like sequences for the early and late viral gene transcription, and interspersed a multitude of host cell DNA- and transcription factor binding sites, the respective promoter/enhancer elements, as well as binding sites for the viral large T-antigen (LTag).
The early viral gene region (EVGR) of ∼2400 bp contains the open reading frames (ORF) of the regulatory proteins small T-antigen (sTag), LTag, and the derivatives T'135, T'136, and T'165. All Tags share a common N-terminal domain, but differ in their C-terminal parts that are derived by alternative splicing of one major pre-mRNA transcript in one direction from the NCCR.
The late viral gene region (LVGR) of ∼2300 bp contains the ORFs of the capsid proteins VP1, VP2, and VP3 which are also derived by splicing of one major transcript, but going in the other direction from the NCCR. In addition, the JCPyV LVGR encodes a small leader protein starting directly adjacent to the NCCR called agnoprotein that has so far only a counterpart in BKPyV and SV40 but not in any of the other known PyVs.
Finally, a gene encoding microRNAs miR-J1B1-3p conserved between JCPyV and BKPyV and some other PyVs has been identified in the distal part of the LTag gene. These microRNAs are generated either from extended LVGR transcripts or by initiation from their own promoter. The microRNAs are predicted to target and downregulate the LTag mRNA and possibly the stress-induced host cell gene ULBP3, a ligand of the NK-cell receptor NKG2D .
Species and genotypes of JCPyV
Comparative analysis of the JCPyV genome shows considerable homology to BKPyV , and more distantly to other human PyVs as revealed by phylogenetic trees (Fig. 3). According to the International Committee on Taxonomy of Viruses and a recent working proposal, the family Polyomaviridae consists of 3 genera called Orthopolyomavirus with the species SV40, BKPyV, and JCPyV; Wukipolyomavirus with the prototype species WUPyV and KIPyV; and Avipolyomavirus with the respective PyVs detected in birds . By current convention, different PyV species must have less than 81% nucleotide identity (see also Ehlers & Wieland, APMIS 2013 in this issue).
Characterization of JCPyV species revealed that there is only one major VP1 serotype, but at least 7 major genotypes. These genotypes, initially distinguished based on restriction site patterns and then on sequences from the intergenic region located between the distal EVGR and LVGR, were preferentially detected in different geographic areas of the world [13, 14]. Initially, 8 major genotypes were distinguished based on coding region polymorphisms from 100 full-length JCPyV sequences . Genotype 5 was classified as being the same as 3, leaving 7 major genotypes, with one to five subtypes. Different ethnic groups around the world were shown to predominantly carry certain subtypes. The degree of the similarity between these subtypes allowed the prediction of ancestral JCPyV genomes and could be correlated with the emergence of the human species in Africa and the subsequent ethnic diversity following pre-historic and historic migration patterns across the world (Fig. 4)[14, 16-18]. For genotype classification, some comments are in order:
First, the VP1 coding sequences of the Asian and South Asian types 2D1, 7C1, and 7C2 are identical and are denoted VP1 consensus. Their distinction therefore requires additional coding sequences from the ORF of VP2/VP3, agnoprotein, and Tag for subtyping.
Second, the European and European American types JCPyV genotype 1A and 1B deviate from the VP1 consensus and show the following amino acids at positions 75R, 117S, 158L, 245K, and 74S, 117S, 126A, respectively .
Third, the Asian and Eurasian genotype 2B bears a 126A change and has been associated with slightly higher risk of PML, whereas the European and European American genotype 4 bearing 134A and 164T appears to have a slightly lower risk of PML .
It is intriguing to speculate that the association of genotype 2B with PML represents a viral virulence determinant, or an immune evasion phenomenon. However, the statistical association might also represent a medical-therapeutic or a diagnostic work-up bias of affected ethnic groups in the developed world. Alternatively, the association might represent behavioral or virological co-factors such as HIV-1 infection. Preliminary data examining the association of HLA class I antigens and PML reported in HIV-infected Caucasian PML patients showed a decreased frequency of HLA-A3, an increased frequency of HLA-B18, and a decreased frequency of HLA-A68 among PML survivors, although statistical significance was not reached after Bonferroni correction for multiple testing . However, two viral markers of PML pathology have been described largely independently of the JCPyV genotype: First, a set of point mutations affecting the sialic acid binding pocket of VP1 [21-24]; and second, diverse unique sequence rearrangements of the NCCR that are thought to be derived from the linear archetype (at-)NCCR [13, 25-28].
The archetype at-NCCR is a conserved linear sequence of blocks denoted A-B-C-D-E-F. In brain and CSF of PML patients, the sequence blocks of the NCCR are rearranged (rr-NCCR) due to partial duplications, deletions, and combinations of both [26, 29]. Given the high diversity and uniqueness of rr-NCCR sequences in PML patients vs the globally conserved linear at-NCCR as majority species in urine around the world, it is generally assumed that rr-NCCRs emerge from at-NCCR during ongoing JCPyV replication [13, 30-32]. However, no clear data are available on the mechanism by which the NCCR rearrangements or the VP1 point mutations are generated and, consequently, the timing and anatomic location of their genesis relative to the onset of PML is debated.
Steps of the JCPyV life cycle
The study of JCPyV life cycle has been hampered by the very narrow human host cell range, which is not well understood, even now, more than 40 years after the report of the initial isolation of the virus by cell culture. Conceptually, at least two levels of restriction have been postulated, the first one acting extracellularly on the cell surface through presence or absence of virus receptors and possibly co-receptors, the second one acting intracellularly at any of the subsequent steps. As detailed below, these steps entail uncoating and nuclear delivery of the JCPyV genome, appropriately orchestrated EVGR expression followed by viral genome replication and LVGR expression, all being integrated via the NCCR, followed by nuclear virion assembly and release of infectious progeny.
Attachment of JCPyV to terminal α2,6-linked sialic acid–bearing structures on the cell surface has been noted early on and is recapitulated by the hemagglutination properties of the virions or VP1-derived virus-like particles (VLPs) . In vitro characterization revealed that the oligosaccharide-bearing receptors mediated internalization via clathrin-coated pits  followed by a Rab5-GTPase-, caveolin-1-, and pH-dependent transport to the caveosome, uncoating steps, and nuclear delivery of the genome . Terminal α2,6-linked sialic acids with virion-binding activity have been identified on oligodendrocytes and astrocytes, on B-lymphocytes in tonsils and spleen, and in kidney and lung tissue consistent with some studies on the tissue distribution of JCPyV DNA [36-38]. Detailed work using glycan arrays identified the highest affinity binding for the pentasaccharide NeuNAc-α2,6-Gal-β1,4-GlcNAc-β1,3-Gal-β1,4-Glc . Crystal structures of the JCPyV-VP1 with and without the pentasaccharide suggested a key–lock interaction with conformational change upon engagement of NeuNAc-α2,6-Gal-β1,4-moiety. Critical amino acids could be identified in the three-dimensional conformation that permitted or, if changed, impeded accommodation of the receptor oligosaccharide in the folded VP1 structure [37, 39]. Although the host cell structure bearing the α2,6-linked sialylated moiety has not been unambiguously identified, glycoproteins and glycolipids are key candidates . By analogy to other viruses, additional or alternative receptor structures might be involved in JCPyV uptake such as the 5-hydroxy-tryptamine-2A serotonin receptor (5HT2AR) . Expression of 5HT2AR in otherwise non-infectable HeLa or HEK293A cells permitted JCPyV uptake which could be blocked by specific drug binding. Although 5HT2AR seems to be sufficient for uptake, infection of cells lacking 5HT2AR indicates that it may not be necessary [41, 42].
The viral life cycle subsequent to JCPyV entry has not been studied in great detail. This is partly due to the fact that archetype at-NCCR JCPyV replicates only very slowly or not at all in vitro. Cultivation of PML variants with rr-NCCR has been only slightly more successful. In fact, Padgett and co-workers reported early on that JCPyV produced only subtle cytopathic effects in the primary human fetal glial (PHFG) cell cultures, and that just a few cells expressed LTag as early as 24-h post infection (hpi), followed by VP1 detection after 48 hpi . The data suggest that JCPyV replication likely follows the general pattern described for SV40 or BKPyV replication. However, both of the latter PyVs yield log-fold increases in supernatant viral loads and significant cytopathic effects after 1–2 weeks [43, 44], whereas this is not the case for JCPyV. The highly heterogeneous JCPyV infection and replication rate in few cells of the PHFG cell culture suggests that synergizing factors are required to overcome the intracellular restriction. For instance, signals of cell activation, cell division, and/or cell differentiation. Recent data suggest that some intracellular restriction might occur through mediators of regulating autophagy such as BAG3 and through induction of apoptosis .
The early viral gene region (EVGR)-encoded Tags are thought to promote viral replication by indirect effects shifting the host cell into a replication-permissive/-promoting state and by direct interaction with viral DNA and viral proteins. JCPyV LTag is a large nuclear protein of 688 aa which, by analogy to the SV40 LTag, is thought to promote JCPyV replication through direct binding to the viral genome and indirectly through interaction with key regulatory proteins of the host cells. Similar to SV40 , LTag contains a DNA-binding domain and a ATPase/helicase activity. Six LTag molecules form two hexameric helicase funnels that directly bind and assist in bidirectional melting of the viral genome starting from the NCCR, recruit host cell DNA binding proteins and the host cell DNA polymerase to the replication fork, and thereby directly promote viral genome replication. LTag also conveys a negative expression feedback of its own transcription by direct interaction with NCCR binding sites, and supports the shift toward LVGR expression. LTag and its splice derivatives T'136, T'136, and T'165 also operate indirectly by subverting cell-cycle control of the retinoblastoma pRB family proteins. The aminoterminal J-domain is shared between all 5 Tags and shows similarities with the cellular DnaJ co-chaperones binding the molecular chaperone Hsc70. Together with the LxCxL motif, this domain binds pRB proteins and releases the E2F transcription factors from the pRB proteins. Similar to the original findings for SV40, LTag binding to the retinoblastoma pRB family proteins modulates the activity and abundance of host cell transcription factors . LTag also counteracts the p53-mediated apoptosis which would be caused by accumulating DNA fragments and metabolic exhaustion, as reviewed elsewhere , and shifts the cell into a proliferative G2/S state by activating ATM- and ATR-mediated pathways . The sTag is a cytoplasmic protein of 172 aa. Its aminoterminal J-domain contains the tripeptide HDP, and one LxCxE motif is involved in binding to pRB family of proteins. The two Zn2+-chelating CxCxxC motifs and one LxCxC motif in the carboxyterminal domain probably mediate inhibition of the protein phosphatase 2A, a negative regulator of the mTOR pathway, through binding and blocking of the other subunits . In addition, binding of sTag and LTag to the enigmatic agnoprotein has been reported. Thereby, JCPyV sTag appears to indirectly contribute to a host cell environment promoting viral genome replication as evidenced experimentally by increasing episomal DpnI-resistant viral DNA .
Issues of JCPyV host cell infection
Various primary human cell culture systems have been studied , including primary urinary epithelial cells , progenitor-derived astrocytes (PDA) , fetal Schwann cells , and oligodendrocytes [55, 56]. However, the poor availability and cumbersome preparation of only limited yield represents a significant obstacle to research, and detailed studies reconstructing convincingly a complete productive JCPyV life cycle are missing. The use of human and non-human cell lines such as IMR-32, HEK293T, SVG-A, Hs685, COS7, HJC-15b, U-87MG, and OWL-586 to overcome these obstacles has had varying success and reproducibility [32, 57-62]. Overall, the results emphasize the critical role of achieving JCPyV EVGR expression together with synergizing host cell signals. Apparently, these effects could be partly mimicked by, or selected for, in cell lines providing a more active LTag of JCPyV or SV40 in trans such as OWL-586, or SVG-A and COS7, respectively, or by relaxing the JCPyV NCCR restriction through JCPyV NCCR hybrids with BKPyV or SV40 . Indeed, most molecular and virological studies were performed by experimentally assembling expression of different viral ORF through transfection, and studying surrogates of intracellular viral genome replication. Having SV40 replication as a model, this approach uncovered and confirmed principle mechanisms and potential interactions, but, as stated, their relative importance remains to be evaluated in a complete JCPyV life cycle.
Animal infections have failed to sustain JCPyV replication and consequently, no experimental model of PML has been established. Instead, the transforming properties of JCPyV were revealed by the induction of brain tumors such as gliomas, medulloblastomas, and astrocytomas . These results documented the oncogenic potential of JCPyV as a result of a non-replicative infection. Apparently, some non-NCCR restricted, accidental expression of the EVGR occurred as demonstrated by LTag expression in some studies. The data suggest that EVGR expression is not sufficient for subsequent LVGR expression and JCPyV progeny production, and that the functional and/or genetic uncoupling of EVGR expression from the tightly controlled or restricted LVGR expression is a potentially oncogenic event. Indeed, genetic uncoupling of EVGR from LVGR expression was evident through chromosomal integration of linearized JCPyV genome parts in these tumors. This was experimentally confirmed in LTag-transgenic mice forming gliomas, astrocytomas, primitive neuroectodermal tumors, schwannomas, and pituitary tumors. As a consequence of LTag expression, increased chromosomal instability was noted as a potential mechanism favoring and selecting further oncogenic events en route to malignancy. In addition, subversion of some cellular signaling and control pathways was discovered, the most prominent ones involving the tumor suppressor neurofibromatosis gene NF2, the insulin receptor substrate (IRS-1)/insulin-like growth factor (IGF-1) pathways, and the β-catenin/WNT1 pathway .
Recently, expression profiling of PHFG cell cultures during differentiation to different lineages was reported to characterize and discriminate important steps toward understanding the intracellular host cell restriction for JCPyV replication . Differentiation to astrocytes with glial fibrillary acidic protein expression was associated with a reactivation of latent JCPyV leading to EVGR and LVGR expression. The study suggests a role for several transcription factors appearing in a time table-like Sequence during differentiation that have been incriminated in earlier cell line experiments and include Spi-B, Pur-α, relA, NF1/CTF, AP1, LEF, and Sp1 . Bioinformatic and experimental mining of these observations can be expected to significantly advance the field of JCPyV biology, to provide clues regarding viral latency and reactivation, and hopefully open new avenues for targeted therapy.
Definition of infection, replication, and disease
The virological and clinical implications of infection, replication, and disease have been discussed previously for BKPyV, which, akin to JCPyV, remains latent after infection and can reactivate viral replication with and without causing disease [65, 66]. The careful use of this distinct terminology is consistent with the current clinical practice for other viruses, e.g., HIV-1 infection, replication, and disease, and applies also to other viruses such as Hepatitis B or cytomegalovirus (CMV) [67-69]. The general considerations of distinguishing infection, replication (multiplication), and disease in microbial pathogenesis are presented in the opening chapter of Mandell's ‘Principles and Practice of Infectious Diseases’  and independently supported by the ‘damage response framework’ .
Virus infection is defined by serological or virological evidence of virus exposure, which includes replicative and non-replicative states. This combination of serological and virological tests has been used in the development of a sensitive and specific antibody test to stratify the risk of PML in multiple sclerosis patients treated with natalizumab .
Virus replication is defined by evidence of virus multiplication as indicated by detection of infectious units, virions, late gene expression such as capsid proteins or their mRNA, or by increasing DNA loads and their detection in non-latency specimens (e.g., CSF, plasma, urine); by cytological or histological detection of cytopathically altered host cells. Today, quantitative PCR and immunohistochemistry are most consistently used to demonstrate replication in the clinics.
Virus disease is defined by evidence of virus-mediated organ damage through histopathology, altered organ function, or impaired organ integrity (proven disease). As tissue samples may not always be available to make the definitive diagnosis of disease, clinical (functional) and radiological abnormalities have been combined with laboratory detection of viral nucleic acid by PCR in relevant, normally negative specimens to define probable disease. The detection of increasing viral antibody may represent an important alternative test to obtain the diagnosis of laboratory-confirmed i.e. virologically confirmed virus-associated disease. Intrathecal antibody production has been used to provide independent laboratory evidence of JCPyV in CNS involvement [73, 74]. A very recent consensus statement by the American Association of Neurology Neuroinfectious Disease Section proposed two separate pathways denoted as histopathologic and clinical PML, and accordingly, classified a CNS disease as certain (=proven), probable, possible, and not PML .
Epidemiology of JCPyV infection
Several seroprevalence studies of healthy human populations detected JCPyV-specific antibodies in 30–70% of healthy individuals. The rates in different age strata indicate that JCPyV exposure occurs during at least two phases: An early phase during childhood reaching approximately 25% in early adolescence followed by a slower second phase during adult life reaching 70% in the sixth decade of life (Fig. 5) [6, 28, 76-79]. By contrast, the seroreactivity to the closely related BKPyV reaches more than 95% during early childhood, followed by slow decline in titers toward the fifth decade of life [6, 28, 76-80]. Thus, JCPyV infection is common in the general human population and transmitted independently of BKPyV, and probably by different routes. Exposure and re-exposure to JCPyV during adult life is associated with an estimated seroconversion rate of approximately 1–2% per year, which is not seen for BKPyV [28, 72].
Healthy blood donors represent bona fide healthy individuals [28, 79], but most other data have been obtained from populations with contact to health-care institutions, some of them using different techniques. A JCPyV GST-VP1-based enzyme immunoassay (EIA) detected an increase in JCPyV seroprevalence from 58.2% in <40-year-olds to 68.5% in >60-year-olds among 424 participants of a skin cancer protection study from Australia . The overall antibody levels were reported as being stable during the 4.5 years of follow-up, and a JCPyV seroconversion rate of 4.4% was reported in line with the other studies. However, fluctuating low-level seroreactivity and even seroreversion was observed . This indicates that cut-off definitions and assay performance with respect to specificity and sensitivity are important issues especially in the weakly reactive and indeterminate zones. This issue was specifically addressed in an international study of 831 multiple sclerosis patients receiving natalizumab using JCPyV-VLPs, which included a 2-step pre-adsorption for confirmation of low reactivity of OD492 nm < 0.250 . The seroprevalence was 53.6% [95% confidence interval (CI) 49.9–57.3%] with an estimated false-negative rate of 2.5% (one-sided upper CI limit of 4.4%). Similar to the Basel study , JCPyV antibodies tended to be higher in male patients and in those older than 50 years . Overall, the antibody levels were stable, yet with an annual seroconversion rate of 2% . In another compilation of 1096 multiple sclerosis patients, JCPyV-specific IgG was detected in 56% (95% CI 53.0–59.0) . In 5 patients testing seronegative according to the assay definition, urinary JCPyV DNA was detected indicating a false-negative rate of 2.7% (95% CI 0.9–6.2) .
Despite using different techniques and assay formats, most studies around the world have confirmed this picture. However, some deviations have been noted in specific populations. Lower seroprevalence rates were observed in multiple sclerosis patients from Britain, Ireland, and Australia compared with patients from the US, Canada, and continental Europe using the same assay . No JCPyV exposure was detected in some isolated tribal populations in South America . By contrast, rapid increases in JCPyV seroprevalence to more than 80% similar to BKPyV were reported in children from Japan [for review, see ]. These differences are impressive and suggest significant early exposure through cultural peculiarities of child care or possibly foods. Given the high JCPyV shedding rate and the detection of JCPyV similar to other viruses in human sewage [85-88], contaminated water, or uncooked or undercooked seafood could be a possibility. Sequence analysis supports the view that JCPyV transmission occurs within and outside of families [89-91]. On the other hand, the median JCPyV seroprevalence of healthy Swiss blood donors at 20–59 years of age was found to be 58%  and a similar rate was seen in Swiss patients with multiple sclerosis , whereas it was 90% in 87 HIV-1 infected patients of the Swiss HIV cohort study using the same VLP-based EIA . The data suggest that risk factors associated with HIV-1 may also be involved in JCPyV transmission (H. H. Hirsch and P. Kardas, unpublished data). A small study from Portugal reported no difference in the overall JCPyV seroprevalence among HIV-1 infected and non-infected persons . However, in the younger age group of 20- to 29-year-olds, the JCPyV seroprevalence was 70% in the HIV-1-infected persons compared to only 50% in the HIV-1-negative group. As male gender is associated with significantly higher rates of JCPyV viruria (see below), and two thirds of the Swiss HIV-1 patients were male, unprotected sex with men shedding JCPyV may be an important cause of exposure during adult life. Together, these data indicate that JCPyV infection is widespread, and is influenced by various environmental, ethnic-cultural, socio-economic, behavioral, and sexual factors related to JCPyV exposure of the general human population. Moreover, previous exposure to BKPyV might mediate some limited immunologic cross-protection as discussed in  (see also below).
Technical aspects of JCPyV antibody assays
The detection of JCPyV-specific antibodies has been accomplished using different techniques including hemagglutination inhibition (HAI), virus neutralization, indirect immunofluorescence of infected or transfected cell cultures, and enzyme immunoassays (EIA) using recombinant JCPyV proteins, all of which are currently based on VP1 presented as monomer, pentamer, or virus-like particle (VLP) . Unlike for BKV [80, 95], there are currently no data comparing JCPyV HAI, anti-VP1, anti-VLP, and anti-LTag responses, except from experimental tumor exposure models . By extrapolation from the BKPyV data, it can be presumed that the strongest JCPyV antibody responses are indeed directed against the viral capsid protein as a result of its abundance, repetitive structure, and boosting upon re-exposure [80, 95, 97]. However, also based on the results from BKPyV, some responses to the viral early proteins, especially to the aminoterminal domain common to sTag and LTag can be expected [80, 95]. In the general population, these anti-N-terminal Tag responses are low or hidden in the background signals [80, 95, 96, 98], but may become detectable after sufficient LTag abundance following high-level PyV replication or from PyV-associated tumor formation [80, 95, 96, 98, 99]. Thus, LTag abundance can result from replicative-cytopathic or from non-replicative oncogenic abundance and might be indicated by increasing antibodies to the N-terminal Tag domain. How this information can be exploited as a marker of risk, recovery, or protection for replicative and non-replicative JCPyV diseases such as PML or possibly cancer remains currently unclear and needs to be investigated.
Regarding the viral capsid antibodies, several studies have correlated HAI, neutralization and VLP EIA [6, 61, 100], but it is becoming clear that the assays are not equivalent with respect to their immunological and clinical implications. Key issues are the choice, preparation, and the purity of JCPyV antigens, the preparation-associated background reactivity, the linear and higher order conformation of the epitopes and their impact on sensitivity and specificity, the identification of cut-offs to define positive and negative test results, the interpretation of indeterminate results in the gray zone, the definition of analytically significant and then clinically relevant quantitative changes, and the detection and interpretation of different antibody classes and subclasses . Cross-reactivity and specificity can be partly addressed by pre-adsorption studies with homologous JCPyV or heterologous PyV antigens from the same recombinant expression and purification protocol. Only limited data are available comparing the results of similar and different antigen preparations and corresponding assays. Therefore, a few considerations are presented addressing the limitations of current assays:
JCPyV virions agglutinate type-0 red blood cells, a property shared by BKPyV but not by the monkey SV40 [6, 77]. The hemagglutination is mediated by binding of the virion to sugar residues on the surface of the red blood cells, which are presumably similar to oligosaccharides of the JCPyV receptor and sensitive to sialidase activity [33, 34, 39]. Accordingly, both JCPyV HAI and neutralization require intact viral particles in three-dimensional conformation that need to be generated and purified from cell cultures, both not available outside of dedicated research laboratories. Hemagglutination and HAI can also be efficiently mediated by VLPs assembled by wild-type JCPyV VP1, but not by VP1 pentamers or monomers.
The source of PyV virions or the recombinant viral antigens may impact the serological background activity as a result of preparative impurities from, e.g., African green monkey cells, E.coli, yeasts, or Sf9 insect cells, and imperfect folding, e.g., E.coli, and may cause increased background and cross-reactivity between PyVs. Given the fact that humans are colonized with E.coli and yeast organisms and form antibodies, the respective recombinant antigens need to be highly purified and appropriate negative controls need to be run to account for anti-bacterial or anti-yeast background reactivity. For Sf9 insect cells or monkey cells, this and conformation issues are presumed to be much less of a problem, but some individuals may have a confounding reactivity.
The antigen conformation is of particular concern for detecting JCPyV capsid antibodies. Recombinant JC-VLPs recapitulate the three-dimensional morphology of PyV virions as shown morphologically by electron microscopy and functionally by mediating hemagglutination and host cell binding [100, 102]. These specific properties cannot, or can only partly, be mimicked by VP1 monomers and pentamers, respectively. Studies comparing JCPyV and BKPyV VP1 vs VLP indicate that the respective antibody binding activities are present and distinguishable in humans, and have suggested a lower sensitivity and a lower specificity for VP1 compared with VLP [80, 103]. There is a greater potential for VP1 monomers and pentamers to detect cross-reacting BKPyV and SV40 antibodies. Part of this activity may correspond to linear epitopes, which are not accessible in the VLP, and some have been characterized by monoclonal antibodies [23, 104]. VP1 cross-reactivity may therefore complicate a routine interpretation of serological results and demand additional pre-adsorption experiments . However, the extent of inhibition that is deemed to be specific is only arbitrarily defined. For one assay, a reduction by greater or equal to 40% has been proposed to be specific . These issues become particularly critical when the overall anti-JCV response is low and the coefficient of variation is higher . Clearly, quality assurance programs and the definition of standardized antibody reference materials are needed as is the reporting of test results in international units.
The functional properties of antibodies and their interpretation as clinical markers of risk or protection are incompletely investigated . Neutralizing antibodies as measured by inhibition of infection are thought to have high specificity and provide protection from infection of host cells and contribute to antigen clearance. The classic assays are not well amenable to JCPyV and they require the generation of infectious virions, the use of susceptible cells, and standardization of the infectious read-out assay. Recent developments may facilitate the measurement of neutralizing activities by pseudo-typed or reporter viruses [108, 109]. By contrast, VLP EIA detects a variety of antibodies that include neutralizing and non-neutralizing binding activities. Comparing HAI and JC-VLP EIA, Hamilton and colleagues demonstrated that there was a good correlation between the respective titers, but that overall the sensitivity was significantly lower for HAI. Although there is only one major JCPyV serotype, point mutations have been detected in PML patients that decrease or abrogate receptor binding to the oligosaccharides . In line with the role of sugar residues for hemagglutination, the mutant VLPs such as L55F or S269T are no longer able to hemagglutinate type-0 erythrocytes (Fig. 6) , but still show an effective EIA activity, which is more than 40% reduced following pre-adsorption (Fig. 7, P. Kardas, F. Weissbach, H. H. Hirsch, unpublished data).
Aspects of JCPyV-specific cellular immunity
JCPyV-specific cellular immune responses are thought to represent the key safeguard against uncontrolled invasive JCPyV replication and disease such as PML. The initial focus has been on CD8+ T cells as the key effectors killing virus-infected host cells when viral epitopes are presented on the cell surface via major histocompatibility complex (MHC) class I molecules. JCPyV-specific CD8+ T cells are found in healthy adults and appear to correlate with protection from PML . However, even in healthy donors, the frequency of JCPyV in the peripheral blood is at least one order of magnitude lower than that found for CMV. In immunocompromised patients, the presence of JCPyV-specific CD8+ T cells correlated with prevention and resolution of PML [111-113]. In a prospective study of PML patients, JCPyV-specific CD8+ T cells were found to have a high predictive value of 87% for the control of PML, whereas their absence had an 82% predictive value for the progression of PML . In PML survivors, JCPyV-specific CD8+ T cells were still detectable several years after the disease suggesting that these responses were either maintained after very high antigenic stimulation or that undetected (re-)stimulation persisted after clearance of the disease . Conversely, CD8+ T-cell responses were low or undetectable in PML non-survivors . JCPyV-specific CD8+ T cells can be found in CSF consistent with local surveillance of the central nervous system . In a dual case–control study matching HIV-infected PML non-survivors with 3 HIV-infected patients of similarly low CD4 cell count, JCPyV-specific T-cell responses were low or undetectable as measured by interferon-γ enzyme linked immunospot assay (EliSpot), whereas the cellular immunity against CMV was preserved in all patients (Fig. 8). Conversely, no difference could be seen between HIV-infected PML survivors and their controls with respect to JCPyV or cytomegalovirus-specific T-cell responses . These data are of interest because they suggest that the decisive parameter for PML progression is the state of JCPyV-specific T-cell repertoire and not that of the overall CD4+ T-cell counts. Thereby, the data partly address the question why not all JCPyV-infected HIV-infected patients with CD4 cell counts as low as 60/μL [115, 116] readily progress to PML. As the interferon-γ EliSpot assay used overlapping peptide pools of 15 amino acids covering JCPyV LTag and VP1, CD4+ T cells and to a lesser extent also CD8+ T cells contributed to this response. As shown for other viruses in HIV and transplant patients, CD4+ T cells are an important marker of the intactness of the cellular immune response by interacting and activating the CD8+ T-cell effector phase. In a recent study, expression of the programmed cell death receptor-1 (PD-1) was found to be increased on JCPyV-specific CD8+ T cells in PML patients suggesting that specific cells are present, but are functionally exhausted . Other data suggest that the presence of CD4+ T cells is necessary for an efficient containment of PML, but CD8+ T-cell responses correlate better with averting disease . In multiple sclerosis patients, a potent cellular immune response against JCPyV has been detected . In addition, no decline in JCPyV-specific T-cell responses was observed over 18 months in the peripheral blood of multiple sclerosis patients treated with natalizumab (Fig. 9) . The data are consistent with the hypothesis that natalizumab prevents homing to sites in the brain, but does not affect the peripheral PBMC responses. The absence of a strong systemic immunosuppressive action was indirectly underlined by the lack of increasing JCPyV DNA detection rates in the peripheral blood . In some patients, the JCPyV-specific T-cell response started to increase after 12 months of natalizumab therapy suggesting that some exposure might have become immunologically visible . Again, it is not clear if this occurred as the result of JCPyV antigen exposure (through replication) in the CNS activation, in the peripheral activation, or in both locations.
The repertoire of viral epitopes recognized by JCPyV-specific CD8+ T cells is not extensively characterized. Most MHC class I epitopes are restricted to the HLA-A*0201 molecule that is found in 20% of the white population [112, 120]. Computer algorithms were used to predict peptide binding to HLA-A*0201 molecule and led to the identification of VP1p100 (ILMWEAVTL) and VP1p36 (SITEVECFL) present in positions 100-108 and 36-44 of the JCPyV capsid protein VP1, respectively [112, 121]. CD8+ T cells specific for VP1p100 and VP1p36 exhibit a memory phenotype and are detectable in a large number of PML survivors, but were rare or absent in PML non-survivors. Furthermore, VP1p100- and VP1p36-specific CD8+ T cells show a potent cytotoxic activity after in vitro expansion and stimulation with VP1p100 or VP1p36 peptides [112, 121]. These data suggest that, despite their very low frequency compared with T cells specific for other viruses, e.g., Epstein–Barr virus (EBV) or CMV, JCPyV VP1p100- and VP1p36-specific CD8+ T cells play a major role in the control of PML [112, 121]. In fact, approximately 75% of healthy individuals have CD8+ T cells recognizing at least one of these two epitopes . VP1p100 is conserved among JCPyV genotypes  and has sequence homology with BKPyV VP1p108 (LLMWEAVTV, position 108–116) . Krymskaya et al. showed that immunization of mice with BKPyV VP1 or in vitro stimulation of PBMC with BKPyV VP1p108 results in the production of CD8+ T cells specific for BKPyV VP1p108 and cross-recognizing JCV VP1p100 . Similarly, JCPyV VP1p36 has been identified as a conserved epitope as it is homologous to BKPyV VP1p44 (AITEVECFL, position 44–52) . Immunodominant epitopes within BKPyV LTag have also been investigated using computer prediction algorithm. BKPyV LTagp579 (LLLIWFRPV, position 579–587) is HLA-A*0201 restricted and located in the p53-binding region of LTag . This epitope shares a complete homology with JCpyV LTagp578, suggesting that CD8+ T cells specific for BKPyV LTagp579 would be able to cross-recognize JCPyV LTagp578 as it has been shown for VP1 . Another conserved epitope is JCPyV LTagp27 (IPVMRKAYL, position 27–35); it is restricted to HLA-B*07 and B*08 molecules and homologous to BKPyV LTagp27 (LPLMRKAYL, position 27–35) .
The JCPyV-specific CD4+ T cells have not been characterized in great detail, but are presumed to play a protective role. This is indirectly supported by individuals with idiopathic CD4+ lymphocytopenia who have been diagnosed with PML, but additional deficits in the adaptive immune response could also be at play [126, 127]. Using computer predictions and in vitro expansion, CD4+ T-cell responses were characterized for specific MHC class II haplotypes . Several immunodominant peptides from different JCPyV proteins were identified, mainly in VP1, but also in VP2 and LTag. The magnitude of the JCPyV-specific CD4+ T-cell response varied for different HLA class II subtypes, and was found to be low or absent for HLA-DRB1*04:01 or HLA-DRB1*01+ . Interestingly, the individuals showed a very pronounced antibody response to JCPyV suggesting that CD4+ TH2-cell help had been effective at one time . It is not clear whether CD4+ TH1-cell function and CD8+ T-cell effectors were also reduced and thereby contributed to higher urinary JCPyV replication.
For a number of persistent virus infections, clinical and virological outcome has been linked to immunogenetic factors including HLA and cytokine polymorphisms. In a first study, differences in HLA allele frequency was reported when 157 HIV-infected Caucasian PML patients were compared to 1752 HIV-infected controls from an American cohort study. Taking HLA-A2 as a reference, a trend for a protective role was suggested for HLA-A3, whereas an increased risk was suggested for HLA-B18, -B41, and -B50. Among Caucasian survivors, HLA-A68, -A31, -B41, and -Cw12 were underrepresented, whereas HLA-B8 and -Cw3 were overrepresented. Different results were obtained for the smaller cohort of HIV-infected patients of African-American ethnicity. Although no statistical significance was reached for any of the analyses after Bonferroni correction for multiple testing, the data are of interest in a re-analysis with larger data sets .
The frequent shedding of JCPyV in the urine of immunocompetent adults is well documented using virus isolation by cell culture, DNA hybridization, or qualitative and quantitative PCR techniques, all indicating JCPyV replication [13, 27, 52, 74, 128-130]. In a comprehensive study of 400 consecutive healthy blood donors from Basel, Switzerland (100 per age decade, 1:1 ratio of females to males), asymptomatic JCPyV viruria was detected in 75 individuals (19%) compared to only 28 (7%) shedding BKPyV . All viruric donors were IgG seropositive for the corresponding PyV. Given the seroprevalence of 58% for JCPyV and 82% for BKPyV, JCPyV replication with asymptomatic viruria occurred in 33% of the JCPyV-infected healthy blood donors compared to 9% of BKPyV seropositive donors . Moreover, the median urine JCPyV load was approximately one order of magnitude higher than the urine BKPyV load (4.64 log10 copies/mL vs 3.51 log10 copies/mL, respectively) . The frequency of JCPyV viruria was higher in men than in women, and increased with age from 12% in the 20–29 years old to 38% in the 50–59 years old donors . The JCPyV NCCR was of archetype architecture in all but one case indicating that rearrangements are not necessary for JCPyV replication in the renourinary compartment [28, 30, 129, 130]. Of note, JCPyV and BKPyV were not detected in plasma using standard diagnostic assays. The differences in epidemiology of these persisting human PyVs suggest again that JCPyV infection is immunologically less tightly controlled than BKPyV in healthy adults, despite evidence for JCPyV-specific immune responses with relatively high antibody titers. Similar data were recently reported for a Portuguese population . JCPyV was not detectable by PCR in oropharyngeal fluids . In a study of 351 immunocompetent outpatients from Japan, JCPyV viruria prevalence increased with age from 44 to 72% using a very sensitive PCR method. In another Japanese study of 120 urology patients, JCPyV viruria was detected in 29% compared to BKPyV viruria of only 4% . JCPyV shedding increased with each age decade reaching 45% in the patients older than 60 years . Conversely, JCPyV viruria was not detected in 120 children from the United States of unknown serological infection status, whereas BKPyV viruria was detected at a low rate of 5% .
The Swiss study of urinary shedding of healthy blood donors at the time of donation is of interest as it excludes potential host factors such as significant health conditions and co-infections such as syphilis, HIV, Hepatitis C virus (HCV), Hepatitis B virus (HBV), pregnancy, inflammation, or liver function abnormalities that have not or only incompletely been tested in other populations. Also, and in contrast to BKPyV viruria, the rate of asymptomatic JCPyV viruria is not increased in immunodeficient patients including HIV-1 infection with low CD4 cell counts [130, 134, 135]. Similarly, comparing 88 HIV-1-infected immunodeficient homosexual men and 88 HIV-1 non-infected controls in the United States demonstrated no difference in JCPyV viruria, being 34% and 37%, respectively, whereas an increase in BKPyV shedding was seen . Thus, decreasing immune control does not affect JCPyV replication, but leads to a pronounced increase in prevalence and loads of renourinary BKPyV replication. The prevalence of JCPyV replication in the renourinary tract does not increase when cellular immunity is systemically impaired as in untreated HIV patients, despite the increased risk of PML and the possible direct activating potential of HIV-1 tat protein on JCPyV replication [32, 136, 137]. Similarly, plasma of HIV-1-infected patients with PML before and at the time of diagnosis was not more frequently positive for JCPyV DNA compared with matched HIV-1-infected controls [101, 138].
The menstrual cycle had no apparent impact on urinary JCPyV shedding in a longitudinal study of daily sampling, covering 36 complete menstrual cycles of 20 healthy immunocompetent women (mean age 25.8 ± 3.1 years), but more than half of the participants reported to use hormonal contraception . Antibodies to JCPyV were found in 6 women (30%), 2 (33%), of whom were shedding JCPyV (one continuously, one in 15% of the samples). Conversely, 95% were BKPyV seropositive and 12% of the urine specimens were intermittently positive in approximately half of the subjects . Besides pointing to false-negative antibody testing results, the data also indicate that point prevalences in cross-sectional studies may underestimate the reactivation rates seen in longitudinal studies.
Pregnancy has been reported to be associated with increased PyV shedding, but this concerned mostly BKPyV and not JCPyV replication . In addition to the hormonal changes during pregnancy, immune responses are changed to accommodate the fetus, which could be viewed as a protected haplo-mismatched tissue graft. In a longitudinal study of 179 healthy pregnant women of mostly African-American ethnicity, 22.3% were found to shed JCPyV at least once during the pregnancy (mean 5.32 log10 copies/mL). In urine samples from pregnant women, individual fluctuations of JCPyV loads were seen throughout gestation, but there were no significant changes in the overall rates, including in the post-partum phase . By contrast, BKPyV viruria rates were higher compared with non-pregnant women, and increased during pregnancy with rates of 13%, 19%, and 39% in the first, second, and third trimester, which were no longer found (0%) post-partum . A study of 19 pregnant women from Italy found rising IgG titers in one and IgM production in two newborn children suggestive of early primary perinatal JCPyV infection . The children also had IgM responses to BKPyV, suggesting the possibility of cross-reactivity or less likely, dual infections. The corresponding three mothers were JCPyV IgG positive, but only two were shedding JCPyV and none of them was shedding BKPyV .
Autoimmune diseases such as multiple sclerosis, inflammatory bowel disease, and rheumatoid arthritis are characterized by an increased reactivity against ‘self’. However, this ‘too much’ of immunity and its associated inflammatory response may also be skewed by gaps in the defense (‘too little’), e.g., against opportunist diseases, which is then further accentuated following immunomodulatory and immunosuppressive therapies. Analyzing 224 multiple sclerosis patients at baseline and at 48 weeks of natalizumab treatment, JCPyV viruria was detected in 58 (26%) and 55 (25%), respectively . JCPyV DNA was not detected in plasma or peripheral blood mononuclear cells (PBMCs), before or after natalizumab treatment, . Similar results were also obtained in smaller cross-sectional and longitudinal studies [92, 142]. The detection of JCPyV DNA in plasma or PBMC in multiple sclerosis patients is rare, occurring in <0.3% of patients, none of whom developed PML, and was absent in patients developing PML . Therefore, unlike BKV viremia in kidney transplantation, JCPyV viremia cannot serve as a specific or sensitive screening marker of viral spread in multiple sclerosis patients [92, 141, 143, 144]. JCPyV viruria was recently examined in 1096 multiple sclerosis patients of the STRATIFY-1 study consisting of 75.7% females and 96.2% therapy-naïve patients . Overall, JCPyV viruria was detected in 20.6 or 36.7% of the seropositives, being higher in males (26.9%) than in females (14.3%) similar to the blood donors. Both, the seroprevalence and the rate of viruria increased with age and were higher in males, but did not correlate with the duration or number of natalizumab infusions (Fig. 10).
Only limited information is available about BKPyV replication of multiple sclerosis patients following natalizumab. In a small pilot study of 57 multiple sclerosis patients from Ireland, BKPyV viruria was present in 8.3% and 22.2% of multiple sclerosis patients before and after a mean of 11 natalizumab infusions, respectively . Two patients with persistent BKPyV viruria developed viremia, one with 17 000 copies/mL clearing after 10 months, but renal dysfunction was not noted .
Thus, JCPyV viruria is a marker of ongoing replication in immunocompetent people and is not readily increased in individuals having various immune alterations. The differences in the immune control of JCPyV and BKPyV are presently not explained, and may reflect differences in the quality, quantity, and accessibility of immune effectors to the site of replicative infection, as discussed earlier . The detection of JCPyV genomes in blood seems unreliable as a screening and monitoring marker. This emphasizes the pathophysiological and diagnostic significance of sampling specific compartments such as CSF or CNS. Also, the analytic sensitivity and specificity must be balanced against the biology of JCPyV latency and persistence, and interpreted accordingly.
Aspects of JCPyV nucleic acid testing
The key techniques for JCPyV nucleic acid testing are real-time PCR, in situ hybridization, and genome sequencing. For body fluids such as urine, plasma, and CSF, real-time PCR represents the first-line diagnostic test. The analytic sensitivity of JCPyV PCR for DNA detection in urine is not a very strong concern unless ambiguous serology results require adjunct testing of, e.g., multiple sclerosis patients. In such cases, further independent sampling should be considered due to the lower point prevalence. The performance of JCPyV PCR is of utmost importance for CSF testing as the sensitivity in histologically confirmed cases ranges from only 70 to 90%, whereas the specificity is considered to be >95% [74, 146]. In principle, the probability of detecting JCPyV DNA in CSF may increase with more extensive viral replication in the brain lesions, but may also be affected by other not well-defined factors such as the clearance rate from the CSF, the size and anatomic location in the CNS, and the recovery of general and specific immune functions. Thus, repeat lumbar puncture and even stereotactic biopsy should be considered, if the initial CSF analysis is negative for JCPyV DNA, but clinical and radiologic suspicion of PML remains high. In HIV-1/AIDS patients with PML receiving combination antiretroviral therapy, or in multiple sclerosis patients discontinuing natalizumab treatment, the likelihood of detecting JCPyV in CSF declines over time in line with the curtailing of JCPyV replication by the recovering immune system [147-149]. Conversely, higher JCPyV copy numbers are associated with shorter survival in untreated HIV-AIDS patients [150, 151]. Following initiation of combination antiretroviral therapy, approximately half of the HIV-1/AIDS patients with PML show disease remission [152, 153] and disappearance of PCR detectable JCPyV DNA in the CSF [74, 151].
For the diagnosis of laboratory-confirmed PML, the analytical sensitivity of the JCPyV PCR should reach 50 copies/mL of CSF in a documented probit analysis . However, most laboratories fail to demonstrate rigorous data in support of this benchmark [154, 155]. The volume of CSF subjected to nucleic acid extraction relative to the elution volume is an important factor of the sensitivity. Typical probit analyses indicate a sensitivity of 3–5 copies per PCR test, which is close to the theoretical limit of one copy. For a limit of detection of 50 copies/mL CSF, a 20-fold concentration of the CSF is required when loading of 5 uL for analysis. Clinicians and laboratory experts need to be aware of these pitfalls, as CSF is frequently a limiting sample in clinical diagnostics, especially when differential diagnosis is broad. Moreover, validated robust and quantitative extraction procedures are needed, as well as controls for PCR inhibition. Timely revision of published target sequences of PCR primers and probes is recommended to identify the need for adaptation as more sequences become available . Confirmatory evidence should be obtained by additional genome targets  and combined with viral genome sequencing.
Aside from PML-affected brain tissue, JCPyV DNA has also been detected in brain tissue of non-PML cases as well as in a variety of extracerebral tissues – including kidney, liver, lung, lymph nodes, spleen, heart, and gastrointestinal tract – by blot hybridization  and nucleic acid amplification methods [158, 159]. In these cases, great care to avoid and test for contamination must be exercised. JCPyV sequencing of the NCCR and the VP1 genomic regions provides important information as to whether or not the detection of JCPyV is of potential pathological significance. Typically, more weight is given to JCPyV detection if the viral sequences in tissue and in blood carry unique, but related NCCR signature rearrangements or characteristic mutations in the VP1 capsid gene [32, 147, 160-162]. However, histological analysis of biopsy or autopsy tissues together with immunohistochemistry for viral proteins remains the gold standard, whereas DNA sequencing, in situ hybridization and electron microscopy are important adjunct tools. The combination of different techniques has been important for characterizing JCPyV-associated nephropathy after kidney transplantation demonstrating the exclusive presence of JCPyV genomes without rearranged, but archetype NCCRs .
JCPyV-associated multifocal leukoencephalo-pathy
The pathology of PML was first described in 1958 as a rare complication of patients with chronic lymphocytic leukemia or Hodgkin's lymphoma, but similar cases were noted as early as the 1930s . PML is a demyelinating disease preferentially affecting the white brain matter and is caused by the cytopathic replication of JCPyV in myelin sheath–producing oligodendrocytes. A Swedish and a US study estimated the incidence of PML in the general population as 0.3 per 100 000 person years (95% CI 0.1–0.6) compared to 1.0 (95% CI 0.3–2.5) for patients with rheumatoid arthritis [165-167]. Similar data were reported for a US study estimating PML in systemic lupus erythematosus, rheumatoid arthritis, and other connective tissue disease as 4, 0.4, and 2 per 100 000 hospital discharges, respectively, whereas the PML incidence was 0.2 per 100 000 hospital discharges in the background population . In another US study, an overall PML incidence rate for several rheumatoid diseases was 0.2 per 100 000 person years .
Higher rates of PML are seen in patients with systemically impaired immune functions such as HIV-AIDS, solid organ transplantation (SOT), and allogeneic hematopoietic stem cell transplantation (HSCT). The incidence of PML in HIV-AIDS has been estimated as 2.4 cases per 1000 patient years in Switzerland which would correspond to 1000-fold increase over the background (Fig. 11) . Recently, similar data have been obtained for Denmark . In heart and lung transplant patients, the incidence rate was estimated as 1.24 per 1000 post-transplant patient years, but the data collection is limited . No cohort data are available in kidney transplant patients, but the general impression is that PML is a rare complication, despite the frequent BK-PyVAN.
For multiple sclerosis patients treated with natalizumab, the incidence has been estimated as 2.13 PML cases per 1000 patients receiving natalizumab for at least 1 month based on 212 confirmed cases among 99 571 patients . Duration of natalizumab treatment for 2 years and more, prior therapy of multiple sclerosis with immunosuppressive drugs, and evidence of JCPyV infection through a positive JCPyV antibody test increased the incidence estimates as high as 11.1 cases per 1000 patients (95% CI 8.3–14.5) .
Rituximab, a depleting monoclonal antibody targeting the CD20+ B cells, has been implicated in PML [171, 172]. In patients with rheumatoid arthritis treated with rituximab, the prevalence of PML was estimated as 1 in 25 000 patients . In a detailed compilation of 57 cases from adverse event registries , PML diagnosis was made after a median of 6 rituximab doses (range from 1 to 28) arising after 16 months of therapy start (median; range from 1 to 90) and 5.5 months after the last dose (median; range from 1 to 66). A fatal outcome was seen in 89.6% occurring 2 months after diagnosis (median; range from 0.4 to 12.2). The data suggest that antibody levels might contribute to JCPyV protection, an interpretation consistent with the increased levels found in HIV-infected PML survivors [93, 101], or that rituximab targets a JCPyV-specific mechanism, e.g., by affecting JCPyV latency in B cells. However, a vast majority of PML patients after rituximab had additional factors associated with an increased risk of PML before the clinical use of rituximab. This included the underlying disease such as chronic lymphatic leukemia and lymphoma, and exposure to corticosteroids (78.9%), chemotherapy, conditioning, and antiproliferative drugs (cyclophosphamide in 73.7%; vinca alkaloids in 57.9%; purine nucleosides in 45.6%; anthracycline in 49.1%). Although this renders a more direct association of rituximab and PML difficult, an increased awareness in this group of patients at risk is clearly warranted and deserves further study.
PML has been described in psoriasis patients treated with efalizumab, a monoclonal antibody blocking the αLβ2-leukointegrin LFA-1 (CD11a) . The clinical manifestations started more than 3 years after efalizumab administration, and both patients died despite plasma exchange 2 and 6 months after diagnosis. Efalizumab was shown in vitro to inhibit transendothelial migration of CD8+ T cells and to reduce T-cell responses , suggesting that indeed impaired CNS surveillance was involved. Both patients had been on other therapies including topical coal tar, salicylates, and sulfadiazine in one, and retinoic acid, PUVA, fumaric acid, and methotrexate in the other case. Exposure to fumarates was recently associated with PML in 2 patients with psoriasis, both of which had pronounced lymphopenia for presumably more than 2 years [174, 175]. Although lymphopenia is frequent in patients treated with fumarates and may even be severe, PML seems to be rare given the more than 180 000 patient years of exposure to fumaderm, a fixed combination of oral dimethyl fumarate, and three monoethyl hydrogen fumarate salts . Thus, pronounced lymphopenia together with fumarate and additional treatments might be operating in concert. In support of the general role of lymphopenia, PML has also been diagnosed in patients with idiopathic CD4+ lymphocytopenia or without (identified) immunological deficits [160, 177-183].
Practically all PML patients are JCPyV seropositive at the time of diagnosis, although JCPyV serology was only investigated in few selected studies of mostly HIV-AIDS or multiple sclerosis patients [72, 73, 93, 101, 170, 184]. The data indicate that parenchymal JCPyV replication and pathology arises despite the presence of a JCPyV-specific humoral immune response. HIV-infected patients with CD4+ T-cell values below 200 cells/μL at the time of the PML diagnosis have a lower survival rate compared with patients with CD4+ T-cell counts above 200 cells/μL [126, 185]. Accordingly, PML is widely regarded as the result of JCPyV reactivation that is not sufficiently controlled by JCPyV-specific cellular immunity, which is mediated by CD4+ helper T cells and CD8+ T-killer cells in healthy JCPyV-seropositive individuals [93, 110, 113, 120]. The role of natural killer cells is not well defined, but, by analogy to other viruses and given the existence of the JCPyV-encoded microRNA miR-J1-p1, which could downregulate the natural killer cell receptor ligand ULBP3, they could play a role during the early or initial phase of primary or secondary replication.
The detailed steps of PML pathogenesis are still a matter of uncertainty as definitive human data are lacking. The reactivation hypothesis is supported by data indicating that virtually every PML case is JCPyV seropositive at the time of diagnosis. Also, prospectively collected sera from HIV-1 patients of the Swiss or US-American cohort studies [93, 101], as well as studies of multiple sclerosis patients receiving treatment with natalizumab, show that patients were JCPyV-seropositive 1–2 years before the onset of PML [72, 186]. However, the time between onset of JCPyV replication in the brain and the accumulation of cytopathic damage resulting in the onset of clinical and radiological abnormalities remains unknown and may well take more than 12 months. Given that PML occurred in 1% of HIV patients in Switzerland during the pre-cART era and that the estimated seroconversion rate is 1–2% per year, there is room for the possibility that some cases of PML are the result of primary infection at a considerable time before diagnosis. Increases in JCPyV antibody titers were noted in some HIV-infected patients more than 6–12 months before the diagnosis of PML [93, 101], and also later observed in multiple sclerosis patients . This increase in JCPyV antibody levels probably represents a humoral boosting response to JCPyV exposure. Irrespective of the underlying mechanism, the data indicate that the adaptive humoral immune system has systemically responded to the JCPyV capsid antigen already months before the clinical diagnosis . If reactivation of JCPyV started locally in the brain, leakage and presentation of antigen and/or cells to the periphery must have occurred significantly ahead of clinical diagnosis. Alternatively, JCPyV reactivation might have occurred in the periphery, or both in the periphery and in the CNS, if JCPyV was latent in both locations and could be systemically reactivated. Interestingly, uncontrolled HIV-1 replication with declining CD4 cell counts takes place systemically and in the CNS. Therefore, HIV-1 replication impairs cellular immune surveillance and can deliver signals activating JCPyV replication through HIV-1 tat targeting NCCR transcripts via a tar-sequence [32, 137, 187-189]. Similarly, immunomodulatory monoclonal antibodies affecting lymphocyte activation and effector functions like natalizumab or efalizumab could be envisaged to deliver JCPyV reactivation signals to cells of JCPyV latency in the periphery, e.g., B cells or bone marrow precursors, while simultaneously decreasing local immune surveillance of the CNS [190-192].
Clearly, JCPyV-specific antibody titer incre-ases do not prevent progression to the clinical disease, although in the Swiss study, this response was associated with increased PML survival . In the latter study, PML survivors had higher JCPyV-specific cellular immune responses than PML non-survivors, emphasizing the role of T-cell responses for the prognosis of PML [93, 110, 113, 120]. Thus, systemic JCPyV-specific antibodies inform about the principle risk of opportunistic reactivation by indicating JCPyV infection, about the more immediate risk by increasing antibody titers in seropositive patients suggesting recent exposure/replication, and about a possibly better survival if coupled to the higher peripheral T-cell responses [93, 119].
The relevant sites of JCPyV latency and persistence that serve as reservoir for PML pathogenesis are not defined. Likewise, neither the time point and the compartment of acquisition of the almost pathognomonic NCCR rearrangements and the VP1 capsid mutations remains known, nor is the compartment and the mechanism of majority species selection identified. However, some principle considerations can be put forward: If JCPyV is transmitted via mucosal surfaces of the oropharyngeal or respiratory tract, then a primary JCPyV viremia must be postulated to reach the renourinary tract. This site is almost invariably colonized with the archetype at-NCCR JCPyV. During primary viremia, a variety of tissues and cells are potentially exposed to JCPyV, permitting infection of non-renal cells and organs including lymphocytes, progenitor–stem, and stroma cells in the bone marrow, tonsils, enteric lymph nodes, and spleen, and possibly also the CNS. The detection of JCPyV genomes by PCR has been reported in many of these sites, although formal evidence of relevant site of JCPyV reactivation and its contribution to PML is lacking [138, 193-196].
Three different hypotheses of PML pathogenesis can be put forward that are not mutually exclusive, but may occur in different clinical settings:
JCPyV reaches the CNS during primary viremia, and reactivation of JCPyV replication is locally censored by specific T-cell effectors. Decreasing JCPyV-specific T-cell surveillance in the CNS permits local replication and cytopathic damage progressing to PML.
JCPyV persists in other cells of the body, i.e., lymphocytes or hematopoietic progenitor cells after primary viremia. JCPyV reactivation in these cells is censored by specific T-cell effectors. Decreasing JCPyV-specific T-cell surveillance permits reactivation and occult secondary viremia leading to colonization of susceptible cells in the CNS, local replication, and cytopathic damage.
JCPyV persists in other cells of the body besides the kidney, i.e., lymphocytes or bone marrow progenitor cells after primary viremia, and reaches the CNS when the infected lymphocytes migrate to the CNS. Reactivation and replication in the absence of specific T-cell control cause colonization of susceptible cells in the CNS, local replication, and cytopathic damage.
Given the blocking of lymphocyte homing by natalizumab, the latter hypothesis is difficult to reconcile with the increased risk of PML, unless other mechanisms or cells are postulated to be involved in JCPyV trafficking. With respect to JCPyV colonization, a recent study reported the detection of JCPyV DNA by PCR in brain of 38% of 28 HIV-positive PML patients, 28% of 18 HIV-positive non-PML patients, and 26% of 19 HIV-negative patients, with tissue viral loads ranging from 377 to 1457 (mean 768) copies/μg total DNA. In situ hybridization confirmed the detection of JCPyV DNA in brain tissues of 3 of 5 HIV-negative patients, and in one case by positive immunohistochemical staining for LTag but not for VP1 protein . JCPyV DNA could also be detected in some extraneural tissues of the HIV-infected patients, which was most frequent in lymph node (34%) and spleen (26%), also bone (18%), but unexpectedly low in kidney (9%). In the HIV-negative individuals, the overall detection rates were lower, but included spleen and bone being positive for JCPyV DNA, but not for LTag protein . Analysis of the NCCR identified mostly rr-NCCR rearrangements in brain tissues of the HIV-positive patients with and without PML as well as some archetype NCCR in the spleen and kidney. Similar results were obtained for the tissues from HIV-negative patients, but at-NCCR was more frequent . In an elegant study using laser capture microdissection of brain tissue from non-PML patients (9 immunocompromised, mean age 56 years; 7 immunocompetent, mean age 77 years), JCPyV LTag DNA was most frequently detected in cortical oligodendrocytes in 78% and 50%, respectively, and cortical astrocytes in 56% and 38%, respectively . When patients with PML were studied, higher kidney JCV DNA load was strongly associated with higher brain JCV DNA load and an inverse relationship was observed between CD4+ T-cell counts and brain JCV load in patients with HIV infection . The data from HIV-negative patients would argue that seeding to the brain occurred without ensuing progression to PML, presumably at the time of primary viremia as the simplest hypothesis. The tissue viral load then increases as immune surveillance wanes in patients at risk for PML. The increased rate of detection rate and viral loads in immunodeficient patients found in some studies suggests that systemic immune surveillance might impact on the renal latency reservoir as well. However, the data in the HIV-uninfected control group do not exclude the possibility of a disease-related ante mortem secondary occult viremia or a role of lymphoid cells in viral trafficking.
The rearrangements of JCPyV NCCRs are highly variable between, and unique within different PML cases, impeding a clear delineation of their role in the pathogenesis of PML and in JCPyV biology. Moreover, in the past, much attention has been given to prototypic strains that have been instrumental in understanding important features of JCPyV, but were isolated by culture from PML cases and possibly underwent additional adaptive changes during in vitro propagation. Direct sequencing of the rr-NCCR after PCR amplification and cloning permitted identification of some patterns (Fig. 12) as discussed below [32, 197].
First, the rearrangements corresponded to duplications of the at-NCCR sequence including partial, multiple, and even linear stretches (tandem repeats), to deletions of the at-NCCR sequence (single, multiple), and complex combinations of both  [for review, please see ].
Second, in rr-NCCR sequences directly from PML cases, deletions typically occurred in the ori-distal part of the NCCR close to the LVGR, whereas duplications occurred in the ori-proximal part of the NCCR close to EVGR TATA-box . JCPyV rr-NCCR deletions most frequently affected parts of, or the continuity of the D-block and less frequently parts of the E and F block. Duplications most frequently concerned parts of, or the continuity of the C-block alone, or in combination with A- and B-block changes. PML variants with complex combinations reflect this basic pattern and appear to be a result of multiple events .
Third, multiple rr-NCCR rearrangements can be found in individual PML cases, but these typically carry a unique rearrangement signature. This is observed when rr-NCCR taken at the same time point, but from different sites like brain, CSF or plasma are compared, or when the same sampling site, i.e., CSF is sampled at two sequential time points  as confirmed in other studies . Sequence analysis from quantitative cloning of JCPyV NCCRs from single-sampling sites indicates the presence of multiple different rr-NCCR sequences carrying a common signature e.g. in brain, CSF and plasma., but with a relative dominance of one majority sequence in that JCPyV genome quasispecies [137, 199]. The majority species in the CSF can rapidly change within 2 days, but will nevertheless carry a rearrangement signature. The data support the hypothesis that ongoing uncontrolled JCPyV replication in the brain is a dynamic process generating a quasispecies of rr-NCCR from the at-NCCR, with a unique founder signature sequence, but varying dominance of rr-NCCR variants as majority species. It cannot be excluded that independent replication foci generate different rr-NCCR in parallel.
Fourth, functional NCCR analysis using a bidirectional reporter assay mimicking the PyV genome organization indicated that, despite the highly variable sequence rearrangements, the majority species rr-NCCR invariably increased EVGR expression compared with the at-NCCR and conferred higher replication rates in PDA cells . The difference between rr-NCCR and at-NCCR EVGR expression could be diminished or abrogated by LTag expression in trans or by HIV-tat expression .
Fifth, the generation of rr-NCCR JCPyV quasispecies in brain and CSF does not indicate whether or not the founder signature sequence was generated from at-NCCR de novo at the time of primary infection/replication or following reactivation in a host with insufficient cellular immunity. The almost constitutive activation of EVGR expression suggests that rearrangements are unlikely to permit latency of JCPyV, i.e., immunologically invisible infection of host cells without EVGR expression. Conversely, the at-NCCR contains sequence elements for HIV-tat , and for the B-cell and transcription factor Spi-B, which are maintained in some, but not all rr-NCCR [32, 137, 191, 200]. This suggests that JCPyV reactivation could be triggered independently of, or synergistically with, normal host cell factors that promote EVGR expression from at-NCCR and/or rr-NCCR. Once replication can proceed from the activated at-NCCR without immunological censoring from T cells, we postulate that errors occur during viral genome replication, probably during recombinant re-circularization, which permit a selection of rearrangements with increased replication fitness . At later stages, rr-NCCRs could emerge that drive JCPyV replication independently of the presumed initial promoting co-factor (‘hit-and-run’), as demonstrated by the loss of the HIV-tat target sequence in an rr-NCCR variant detected in HIV-AIDS PML .
The diagnosis of PML can be viewed as a late clinical and virological state after several generations of cytopathic JCPyV replication has occurred. The loss of myelin-producing oligodendrocytes leads to deficits in the affected neuronal areas with corresponding clinical symptoms and signs, and their radiological correlates. PML lesions are often asymmetric and focal in the initial stages and then show a dramatic subcortical expansion, confluence, and even contralateral spread along the corpus callosum. The frequently asymmetric single-site presentation of PML lesions suggests that the efficacy of JCPyV infection and/or JCPyV replication is variable and possibly depends on additional local factors. It is presently unclear, whether the focality of PML lesions reflects stochastic differences at the time of CNS seeding during primary or secondary viremia, or differences during homing of lymphocyte trafficking, or an as yet undefined heterogeneity of other local factors promoting infection and subsequent JCPyV replication. It is also unclear how oligodendrocytes are infected following presumed vascular or lymphocyte seeding [38, 196, 201-203], and whether it occurs directly or via other cells as intermediates of traffic or even infection. Astrocytes and neurons have been deemed as being largely non-permissive to JCPyV replication, but the activated phenotype of astrocytes and evidence for LTag expression in granule cell neuron suggest that some viral entry and EVGR expression might occur [38, 196, 204].
The clinical presentation of PML is variable and includes progressive focal neurological deficits of mostly motor, cognitive, and visual functions (Fig. 13). Ataxia can be a dominant clinical sign with frank inability to walk, talk, dress, or eat. Epilepsy has been rare in the initial clinical description of PML in HIV-AIDS, but may be more frequent in PML lesions with more cortical location  and in patients with radiological signs of inflammation as a result of local immune responses.
Magnetic resonance imaging (MRI) with fluid-attenuated inversion recovery (FLAIR) and/or T2-weighted enhancement has become the key non-invasive diagnostic method to identify patients with possible PML (clinical symptoms and signs and compatible lesions on MRI), and to document areas of involvement and to provide radiological correlates of the course of the lesions [206, 207]. The most typical locations are in the subcortical white matter with sharp demarcation toward the gray and diffuse borders towards the white matter, but in approximately half of the cases, cerebellum, thalamus, and basal ganglia are affected with varying proportion of gray matter. Typically, hyperintense signals in single, multiple, and/or confluent subcortical areas are seen in FLAIR and T2-weighted studies, being hypointense signals in T1-weighted studies . The absence of contrast enhancement is typical for PML in HIV-AIDS and may reflect the low overall systemic immunodeficiency. Conversely, contrast enhancement is more frequently seen in PML patients treated with natalizumab in whom immune surveillance of the CNS is locally reduced, but probably not completely abrogated allowing for T-cell leakage and effector responses at the site of damage during the course of disease [206, 207]. Activity in the vicinity of Virchow–Robin spaces may give rise to punctate signals . Accordingly, swelling and intracranial mass effects are not seen unless CNS immune surveillance rapidly recovers leading to a strong inflammatory response called immune reconstitution inflammatory syndrome (IRIS) [208-211]. Other adjunct techniques such as diffusion-weighted imaging may indicate older lesions with lower central signal intensity, and techniques using different markers may increase sensitivity and specificity of the radiologic studies, and help in the differential diagnosis of other entities such as relapsing multiple sclerosis. Computed tomography may show hypointense non-enhancing larger lesions, but is generally considered to be too insensitive, especially for early lesions.
The diagnosis of proven PML (histologically confirmed PML) requires brain tissue demonstrating areas of demyelination in the white matter, compatible cytopathic alterations (enlarged oligodendrocytes with intranuclear inclusions; giant, partly multinucleated astrocytes; foamy macrophages; and debris of cellular and nuclear origin), and specific JCPyV involvement (immunohistochemistry for LTag or VP1; DNA by in situ hybridization). The detection of JCPyV by PCR in cerebrospinal fluid of an immunocompromised patient with (multi-)focal neurological deficits and corresponding radiological findings is generally accepted as probable PML, also termed as virologically confirmed or laboratory-confirmed PML [138, 212]. The diagnosis of possible PML is made when JCPyV is undetectable in the CSF by a sensitive PCR detecting at least 50 copies/mL obtained from a patient with typical risk profile and compatible clinical and radiological signs. For cases with CSF-negative PCR results, the detection of increasing intrathecal antibodies may represent an adjunct diagnostic test as well as the detection of JCPyV genome loads in brain biopsy together with demonstration of rr-NCCR sequences .
The clinical outcome of PML varies widely and is thought to reflect the severity and location of the lesions, the underlying condition, and the ability to mount a JCPyV-specific cellular immune response curtailing JCPyV replication. One of the largest systematic observational studies has been the Swiss HIV Cohort Study documenting during 20 years 226 PML cases in HIV-AIDS patients . For 186 patients, sufficient data sets were available, and for 159 patients, the diagnosis was made before death and allowed for comparison of characteristics and outcome of PML patients before (n = 89) and after introduction of cART in 1996 (n = 70). Overall, 97 patients died within 1 year after diagnosis showing a median survival time of 90 days (Interquartile range, IQR 53–312). The median time to PML-attributable death was 71 days (IQR 44–140). With introduction of cART in 1996 in Switzerland, the prevalence of PML declined 10-fold from approximately 1–0.1% . The PML-attributable mortality decreased from 82 cases to 38 cases per 100 person years. Of note, in both eras before and after introducing cART, the patients had similar immunologic and virologic parameters showing median CD4 cell counts of 60 (IQR 20 – 140) vs 71 (IQR 31 – 132), respectively, median HIV-1 RNA log10 loads of 4.9 (IQR 4.1–5.5) vs 4.9 (IQR 3.3–5.4), respectively . In multivariate models, cART was the only factor associated with lower PML-attributable mortality (Fig. 14), whereas all-cause mortality was dependent on both, CD4 cell counts >100/μL and cART . Undetectable JCPyV-specific T-cell response in the peripheral blood and a high or persisting JCPyV load in CSF have been associated with poor prognosis [113, 151, 185, 213]. In a recent review of published and unpublished institutional PML in 44 SOT and 25 HSCT patients, the median survival time was 6.4 months in SOT vs 19.5 months in HSCT patients. The survival beyond 1 year was 56% and the overall case fatality was 84% .
The treatment of PML aims at regaining immune control over JCPyV replication. For transplant patients, this approach requires reduction in immunosuppressive treatment, which is limited by immune reactions causing graft rejection and loss of organ functions. In the case of kidney transplantation, return to hemodialysis remains a viable option, and discontinuation of immunosuppression may be indicated for a successful outcome . In the case of SOT without artificial organ substitution, reducing immunosuppression may not be feasible to allow for a timely recovery of JCPyV-specific immune control. Antiviral treatments with cidofovir and cytosine arabinoside have not been successful. In the case of HSCT, specific immune recovery may be slow and reduction or discontinuation of immunosuppressive medication may precipitate or aggravate clinically severe, life-threatening graft-versus-host disease.
For PML in HIV-AIDS, the start of cART can significantly improve the overall outcome, but still half of the patients succumb to the disease or suffer from significant neurological impairment. Apparently, JCPyV-specific immune reconstitution is too inefficient, too slow, or too late. This is in part due to the fact that there is currently no validated surrogate marker for screening or identifying patients at risk early, before significant damage has occurred. It is currently widely accepted that plasma JCPyV loads are neither specific nor sensitive for PML, and that frequent CSF sampling is not sensitive enough to justify such an invasive procedure for screening purposes. Consequently, a high index of suspicion and careful, frequent clinical and radiological studies are currently the only alternative (Fig. 15). In the case of multiple sclerosis patients receiving natalizumab, this has resulted in an earlier identification of probable and proven PML cases, sometimes even without clinical symptoms , and permitted the prompt discontinuation of the drug and plasma exchange. On the other hand, immune reconstitution may have significant side effects including IRIS , which may aggravate the primary clinical PML presentation including ‘unmasking’ as described in HIV-infected and also multiple sclerosis patients [210, 215], and/or lead to exacerbation of the underlying condition such as autoimmunity, transplantation, or malignancy [211, 216]. In fact, PML-IRIS seems to be a more frequent complication in multiple sclerosis patients after discontinuing natalizumab consistent with the hypothesis that the gates for lymphocyte homing to the brain are then wide open. Plasma exchange to accelerate natalizumab clearance may aggravate IRIS . A recent study suggests that the earlier IRIS occurs, the poorer the outcome with only limited response to corticosteroids . Given the competing risks of PML progression and life-threatening IRIS, the use of corticosteroids is advocated to prevent or treat threatening herniation as the result of inflammatory mass effects. However, uncritical routine use or extended high-dose use of corticosteroids may inhibit immune recovery with fatal outcome . In addition, activation of JCPyV replication from glucocorticoid response elements in the NCCR might further accelerate JCPyV replication and cytopathic damage.
As an alternative in these high-risk patients with PML-IRIS, we have proposed the use of high-dose intravenous immunoglobulins, not primarily for their neutralizing and opsonizing properties, but because of their immunomodulatory activity dampening IRIS. The corticosteroid-sparing activity of high-dose intravenous immunoglobulins is used in idiopathic thrombocytopenia and some cases of acute rejection in SOT. The blood–brain and blood–CSF barriers limit delivery, but most likely, the barrier is significantly impaired at the sites of IRIS. As it is unlikely that intravenous immunoglobulins can substitute corticosteroids in severe and progressive IRIS, early administration may be beneficial as discussed in a recent case report . Given the paucity of PML, and the lack of effective alternatives, the high costs of high-dose intravenous immunoglobulins can probably be justified, but would require a prospective clinical trial. Adoptive transfer of JCPyV-specific T cells has been performed as an adjunct treatment in a child after allogeneic HSCT . This immunologic treatment was well tolerated and no recurrence of PML was noted. Although this indicates principle feasibility and tolerability, the risk of IRIS should not be underestimated especially in patients with florid PML, and should be addressed in clinical study protocols.
The role of antiviral therapies for PML is undefined due to the lack of clinically active agents. Regardless of whether JCPyV-specific immune control prevails as the ultimate therapeutic goal, antiviral agents must be considered as beneficial by reducing JCPyV-mediated cytopathic damage, extending the time window for immunologic recovery, and reducing the risk of IRIS. Moreover, a well-tolerated, effective, and affordable antiviral drug could be used in a prophylactic setting to prevent the onset of PML. Several reports address the potential of the antivirals cytarabine  and cidofovir [152, 219], and uptake blockers mirtazapine, chlorpromazine, and mefloquine alone or in combination, with varying success, typically in uncontrolled case studies [178, 220-224]. Currently, the lipid hexadecyloxypropyl derivative of cidofovir called CMX001 has the highest activity in an in vitro study of JCPyV replication in PDA cells . In a case of PML in idiopathic CD4 lymphocytopenia, the cytokine IL-7 has been used successfully in combination with CMX001, but further trials are needed . Overall, no randomized clinical trials are available that indicate that one or the other drug alone or in combination is effective. Given the poor prognosis, rareness, and diversity of PML cases, large international multicenter studies are urgently needed. Finally, monoclonal antibodies to JCPyV virions are potentially of interest to shift the balance towards interruption of JCPyV replication and more efficient immune control. Such monoclonal antibody should exhibit neutralizing and opsonizing activity for the different VP1 subtypes as well as for the characteristic PML VP1 point mutants [23, 24].
Other JCPyV-associated central nervous system disorders
JCPyV-mediated granule cell neuronopathy describes the cytopathic replication of JCPyV in granule cell neurons of the cerebellum [227, 228]. Clinically, this entity is characterized by a cerebellar syndrome with ataxia and progressive cerebellar atrophy. Radiologically, cerebellar lesions show atrophic areas without white matter involvement . Virologically, JCPyV can be detected in the CSF, and viral genome sequences revealed C-terminal frame-shift mutations and truncations of the VP1 capsid protein. If the VP1 truncations can be confirmed as a specific virus determinant associated with or selected in this pathology in independent studies, a convincing hypothesis and its experimental testing still needs to be put forward [162, 229]. In another report, no major alterations of the C-terminal VP1 were found, except for amino acid changes at methionine-339 to isoleucine, and lysine-345 to arginine . The NCCR was rearranged with a partial deletion of the D- and B-block, whereas the CSF variant only showed a partial D-block deletion as signature rearrangement . The histopathology showed LTag expression in neuronal cells staining positive for MAP-2 in the granule cell layer [228, 231]. Granule cell neuronopathy has been proposed as an independent pathological entity in HIV-infected and non-infected patients . The lesions frequently coexist with the PML in the white matter of the cerebellum as well as in the cerebrum. As replicative pathology, the proposed treatment aims at regaining immune control of JCPyV replication, but the atrophic areas persist.
JCPyV-associated encephalopathy describes a second entity of cytopathic JCPyV replication in the gray matter, which targets cortical pyramidal neurons . This entity was first described in an HIV-negative woman with a history of lung cancer, who suffered from progressive cognitive deficits, aphasia, and seizures before death . MRI revealed multiple non-enhancing lesions in the gray matter, which expanded to subcortical areas. The CSF and autopsy tissues were positive for JCPyV by PCR. Sequencing of the viral genome identified a rr-NCCR with several smaller duplications and deletions. Moreover, a variant having a 143 bp deletion of the agnoprotein ORF coexisted with a variant bearing an intact agnoprotein ORF . The role of these variants with respect to pyramidal neuron cell tropism and replicative capacity remains unclear. The coexistence with agnoprotein-deleted and non-deleted variants emphasizes the JCPyV rr-NCCR quasispecies concept. It also suggests the possibility that potentially lost viral functions can be rescued by dual infections as discussed for BKPyV rr-NCCR with agnoprotein deletions in kidney transplants with PyVAN .
JCPyV-associated meningitis or encephalitis has been described in cases of patients with typical clinical signs of meningeal inflammation (neck stiffness, headache), fever, and the detection of JCPyV in the CSF [235-237]. Clinically, no focal neurologic deficits characteristic of PML are found, and MRI shows no evidence of subcortical T2-enhancing white matter lesions. No histopathology studies have been reported from these cases. The diagnosis is suggested based on the clinical symptoms and signs of meningitis, the laboratory work-up demonstrating JCPyV in the CSF and the exclusion of other etiologies.
Nephropathy due to JCPyV (JCPyVAN) has been recognized as a rare complication in less than 1% of kidney transplant patients, but can cause severe graft dysfunction and progressive loss [163, 238]. The clinical presentation is inconspicuous, but in advanced cases, the renal allograft function declines from the post-transplant baseline. Virologically, very high urine JCPyV loads of >7 log10 copies/mL have to be differentiated from levels found in healthy populations , and are typically found together with the urinary shedding of ‘decoy cells’, cytopathic epithelial cells with characteristic intranuclear inclusions. By definition, a role of BKPyV is excluded by undetectable or very low urine BKPyV loads. Of course, analytical issues including BKPyV assay inhibition and target mutations need to be ruled by confirmatory tests. Testing of blood or plasma is unreliable as JCPyV DNA is frequently low or undetectable yielding poor negative and positive predictive values . Accordingly, no general screening recommendation has been given for JCPyV replication, and the diagnosis relies on a high index of suspicion in BKPyV-negative cases. Proven JCPyVAN is diagnosed histopathologically by immunohistochemical detection of LTag in renal tubular epithelial cells using the cross-reactive SV40-LTag antibody also used for BKPyVAN. No morphological differences compared with BKPyVAN have been described. In a prospective study performing biopsies for persisting decoy cell shedding or declining allograft function, six cases of JCPyVAN were identified among 28 patients having high-level JCPyV viruria . In these cases, cytopathic changes were less pronounced or scarce, but patches of chronic inflammation and fibrosis were prominent. Although no JCPyV-specific immunohistochemistry was performed, e.g., for VP1 [241, 242], the diagnosis was confirmed by the exclusive detection of JCPyV DNA in three of the six biopsy cases showing up to 1000 copies/diploid cell equivalent. The NCCR sequence showed no rearrangements indicating that the pathology of JCPyVAN occurs without rr-NCCR. This is similar to BKPyVAN, which arises with at-NCCR BKPyV that are, however, subsequently replaced by rr-NCCR BKPyV variants as majority species if uncontrolled virus replication has been ongoing for some time . The factors permitting relatively high-level replication of the JCPyV archetype in the urinary tract of immunocompetent individuals and causing invasive renal disease in some kidney transplant and HIV patients are currently undefined.
The treatment of JCPyVAN by reducing immunosuppression can stabilize allograft function during follow-up and lead to clearance of histological involvement . The risk factors for JCPyVAN are not defined. The entity affected deceased and living donor grafts, but has been reported in slightly more male recipients on tacrolimus–mycophenolate–prednisone combinations [163, 238, 242, 244, 245]. In a recent study report, JCPyV viruria occurred earlier in recipients of grafts from donors with higher JCPyV antibody levels . Co-detection of JCPyV and BKPyV viruria was lower than expected suggesting an inhibitory virological and immunological interaction . In HIV-AIDS patients, JCPyV replication has also been reported and should be considered in cases of otherwise unexplained renal failure .
The transforming potential of JCPyV is well documented in different cell biological situations, which provide biological plausibility and raise the question about the role of JCPyV in human malignancies. A number of human malignancies have been associated with JCPyV including oligodendroglioma, astrocytoma medulloblastoma, ependymoma, and glioblastoma [189, 248-251] and non-Hodgkin lymphoma [252, 253], as well as colorectal carcinoma, and gastric and anal cancer [254-260].
As discussed elsewhere in more detail , the oncogenic contribution of a ubiquitous polyomavirus like JCPyV is difficult to ascertain. The careful studies regarding the recently discovered Merkel cell carcinoma virus provide good examples of how to approach this question. The virus could be that of a ‘driver’ contributing actively to the clinical and pathological properties of a given malignancy. In that case, oncogenic JCPyV-encoded gene products needed to be expressed in all the cancer cells and a knockdown of that viral gene function would cause at least a partial reversion. In the alternative ‘hit-and-run’ scenario, JCPyV gene products would contribute only to initial steps of the oncogenic transformation, but subsequent progression steps of the malignancy are then formally independent. In that case, knockdown of JCPyV gene products would not affect the malignant phenotype, if expression of the viral gene was not lost during malignant progression. Both the ‘driver’ and the ‘hit-and-run’ situations need to be distinguished from the ‘passenger’ role, where virus persists in cancer cells because of a suitable primary or secondary host cell specificity, but does not actually contribute to the malignancy. Finally, in the ‘innocent by-stander’ situation, JCPyV would replicate in unrelated cells in the vicinity of the malignancy. Accordingly, the mere detection or non-detection of viral genomes is neither necessary nor sufficient to prove or rule out a contribution to malignancy. Given the somewhat mixed level of evidence regarding JCPyV and cancer, a stringent evaluation algorithm seems to be required. This could be achieved for instance by a step-wise approach focusing on criteria identifying cancer ‘driver’ role of JCPyV by biologic-mechanistic plausibility; epidemiologic association in specific cancer populations exposed to JCPyV, and diagnostic confirmation through the presence of the viral genome, relevant viral mRNA transcripts, and viral early proteins, and oncogenic read-outs, e.g., p53 stabilization, IRS-1 or WNT1/catenin signaling. However, independent confirmation in a blinded laboratory setting should be incorporated .
JCPyV-associated oligodendroglioma has been described in non-PML patients [248, 250]. Biological plausibility could be postulated as this cell type can be infected by JCPyV provided that primary or secondary seeding of JCPyV to the brain occurred as discussed. The relative immune privilege of this sanctuary location would account for some degree of escape from cellular immunity, which should be even higher in immunocompromised patients. Epidemiological data are not available, but JCPyV infection is frequent as derived from the general seroprevalence data. In some cases, LTag expression was demonstrated by staining and western blotting [189, 248]. JCPyV DNA was detected, but the NCCR corresponded to the MAD-4 strain rather than a unique rr-NCCR . Given the emerging concepts about NCCR rearrangements, there are questions about the probability that such a rearrangement would re-occur independently or that the primary JCPyV infection occurred with a Mad-4 strain.
JCPyV-associated oligodendroglioma has been reported in patients with PML after chemotherapy for hematological cancers. The presence of PML provides a high local JCPyV load, which could increase the probability of genetic alterations in the NCCR, infection of non-permissive cells, or accidental chromosomal integration of the viral genome that permit EVGR expression with sTag and/or LTag, but aborting cytolytic LVGR. These events could be supported by an inherited genetic predisposition for malignancy or acquired somatic genetic alterations as a result of the genetic instability due to LTag expression, radiation, or chemotherapy. However, given the rareness of PML, tumor formation would be missed without a systematic radiological and histopathologic survey, even if it occurred at a relative high rate of 10% of PML cases. Moreover, the poor prognosis of PML may be a competing risk with the onset and outgrowth of associated JCPyV cancers. Conversely, recovery of the immune control clearing PML would also increase the opportunity of clearing LTag-expressing cancers, but not of ‘hit-and-run’ progeny.
For JCPyV and non-Hodgkin lymphoma, epidemiologic data are presented in a 1:2 matched nested case–control study of blood donors . The presence of antibodies to JCPyV alone was not found to be associated with the subsequent onset of 170 cases of lymphoma (odds ratio 0.83, 95% CI 0.56–1.23). This result was not changed after adjustment for EBV seropositivity, induction period, or lymphoma subtype. However, among blood donors showing an increase in JCPyV antibody levels in the sampling interval from 1974 to 1989, the risk of lymphoma was 4.59-fold increased (95% CI 1.3–16.25; p = 0.02). The data are intriguing and illustrate that a combination of epidemiologic, diagnostic, and mechanistic studies are needed to investigate the transforming potential of JCPyV or other PyVs in humans. JCPyV has been detected in 5 of 16 EBV-positive primary B-cell lymphomas of the CNS . In these lymphomas, JCPyV LTag and EBV latent membrane protein-1 (LMP-1) were co-expressed, but LVGR expression was not detected as evidenced by VP1. Although these data suggest a potential role of JCPyV, the 11 JCPyV-negative, EBV-positive cases argue that JCPyV is not, or no longer (‘hit-and-run’) necessary, and further data are required to distinguish JCPyV as ‘passenger’ rather than as ‘driver’ for this malignancy.
For gastrointestinal cancers, a recent study detected JCPyV DNA in 26% of 61 primary gastric carcinoma samples, but only 6% of 59 paired non-cancer mucosa controls . Patient age and the methylation index were the only independent factors associated with JCPyV detection. Interestingly, JCPyV positivity was associated with a trend toward improved survival inviting some speculations about the presence of JCPyV in an earlier cancer stage or better immunologic discrimination. Together, the data indicate the challenge to convincingly link a ubiquitous infectious agent mechanistically and epidemiologically to cancer, and to translate this information into rational diagnostics and therapeutic decisions in clinical practice.
Since the discovery of JCPyV almost half a century ago, significant progress has been made in understanding the epidemiology and biology of this rather inconspicuous virus. These advances can be ascribed to the efforts of researchers in basic and clinical science to identify factors that transform JCPyV from a common companion to an aggressive pathogen, causing a rare, but devastating disease. Clearly, there are still significant gaps in our knowledge. As PML arises in different populations of immunologically vulnerable patients, careful study of the co-factors may allow identification of shared vs unique mechanisms, and thereby help in unraveling critical steps of the pathogenesis. Although some compelling concepts have been proposed, their critical review and experimental challenge is still partially pending. Hopefully, future studies will generate better evidence and can be applied to dissect other replicative and non-replicative JCPyV pathologies to better diagnose, treat, and prevent JCPyV-associated diseases.
We wish to thank Christine H. Rinaldo and Garth Tylden for critical reading of the manuscript and helpful comments.