PML's 50^th birthday- New Insights into Pathogenesis

Progressive multifocal leukoencephalopathy (PML) is an opportunistic demyelinating disease of the central nervous system caused by JC polyomavirus, that was initially described by Astrom, Mancall and Richardson half a century ago (1). PML develops when JC virus productively infects oligodendrocytes leading to their lytic destruction, possibly through mechanisms involving apoptosis, with resulting demyelination. Cases of PML emerging during treatment with several immunomodulatory agents of great therapeutic promise for treatment of immune-mediated neurological and non-neurological disease, including natalizumab (Tysabri), rituximab (Rituxan), and mycophenolate mofetil (CellCept) (see http://www.fda.gov/medwatch/safety/2008/safety08.htm), has re-focused emphasis on understanding the pathogenesis of these devastating and currently untreatable neurological disorder.

JC virus itself is a ubiquitous agent, and seroepidemiological studies suggest that the majority of the population has been infected before reaching adult life (2). Primary infection occurs in childhood and is asymptomatic or subclinical. Following primary infection virus becomes latent in the host, persisting unobtrusively in the overwhelming majority of individuals for their lifetime. What are the factors that allow this normally innocuous and apathogenic virus to produce devastating neurological disease in the form of PML?

A critical determinant for the development of PML is the requirement for lytic infection of oligodendrocytes. How does virus reach these cells? The prevailing theory of PML pathogenesis suggests that after primary infection JC virus becomes latent in a variety of host cells likely including the tubular epithelial cells of the kidney, B-cell progenitor cells in the bone marrow, circulating B-cells and tonsillar stromal cells (3). PML is essentially an opportunistic infection and well-documented cases in immunocompetent individuals are vanishingly rare. Individuals with chronically impaired cell-mediated immunity are at particular risk of developing PML. The presence in blood or CSF of JCV-specific MHC-I restricted CD8+ T-cells specific for epitopes on the viral capsid protein VP1 is associated with a reduced risk for development of PML and an improved prognosis if disease occurs, suggesting that these cells play a critical role in the host's immune response against JCV(4). The dramatic association of disease with HIV infection and its occurrence in individuals with CD4 lymphopenia in the absence of HIV, suggests that CD4+ T-cells also play an important role in controlling infection (5). Conversely, humoral immunity seems less important than cellular immunity in control of PML, as individuals with isolated depression of humoral immunity but preserved cellular immune responses only rarely develop PML, and disease occurs and progresses despite a robust anti-JCV antibody response in both the blood and CSF.

In the conventional model of PML pathogenesis, depression in cell-mediated immunity allows for reactivation of virus, which then enters the bloodstream (viremia) and subsequently disseminates to the CNS where it productively infects oligodendrocytes via the serotonergic 5HT2a receptor (6). Infection of astrocytes is abortive, although these cells are transformed into giant bizarre forms that are one of the characteristic neuropathological hallmarks of PML. Neurons are not infected in PML, although some JCV isolates with deletions in the VP1 cell attachment protein can produce a syndrome of cerebellar dysfunction resulting from infection and injury of cerebellar granule cell neurons (7). JCV DNA has been amplified by PCR from B-cell, T-cell, monocyte, granulocyte, and even plasma fractions of HIV infected individuals with or without PML but is not generally detected in blood of immunocompetent HIV-negative individuals (8).

There are important implications of the conventional model of PML pathogenesis for diagnosis and therapy. If this model of JCV pathogenesis is correct, virus would be expected to appear in the blood and/or CSF prior to its appearance in CNS tissue, and therefore before clinical and/or radiographic signs of PML were apparent. A sensitive method for early detection of virus in blood or CSF might provide a warning interval that could allow for discontinuation of predisposing immunomodulatory therapies or specific antiviral treatments as these become available.

An alternate theory of the pathogenesis of PML suggests that viral latency occurs in the CNS, and is established at the time of primary infection. This theory gains credence from early studies suggesting that JCV DNA could be detected by in situ hybridization and PCR amplification in the brains of apparently normal individuals (9-11). This theory also addresses another central observation in JCV biology. The genome of JCV contains a noncoding control region (NCCR). In patients with PML, the NCCR of JCV isolated from the brain typically has a rearranged form (“Mad”) that differs from the NCCR (“archetype”) typically found when virus is isolated from the kidney and other extraneural sites (12). The conventional model of PML pathogenesis presumes that virus reactivates in extra-neural sites, and then undergoes genetic rearrangement from the NCCR_Arch to the NCCR_Mad genotype before it becomes neuroinvasive. In the alternate model, it is assumed that viruses with both Mad and archetype NCCRs circulate in the general population, but that only the NCCR_Mad viruses can achieve CNS latency and/or reactivate from oligodendrocytes. In this model, reactivation is presumed to occur directly in oligodendrocytes or nearby cells with later dissemination of virus to CSF and/or blood. In this model, the key site of immune surveillance is not the periphery but the CNS itself, with depressed cellular immunity facilitating direct CNS reactivation of JCV rather than reactivation at an extraneural site with subsequent spread to the CNS. Obviously in such a model the detection of virus from extraneural sites such as blood or CSF would not occur consistently until after disease was already established in the CNS and as a result would be of limited predictive or therapeutic utility.

The studies by Khalili and colleagues reported in this issue of the Annals (13) have important implications for these alternate theories of JCV pathogenesis. The authors' utilized PCR to amplify JCV DNA corresponding to specific JCV genome regions encoding T-antigen, the major viral capsid proteins (VP1), the viral agnoprotein, and the NCCR, from postmortem brain tissue of seven HIV-negative individuals without PML. T-antigen DNA was found in 5/7 cases, VP1 DNA in 4/7, agnoprotein in 6/7 cases and NCCR DNA in 3/7 cases. In the one patient (Case 2) from which the NCCR could be sequenced it was of the neurotropic Mad4 genotype. Every one of the seven patients examined had JCV DNA corresponding to at least one part of the JCV genome found in at least one region of their brain, and one patient (case 3) had DNA from every portion of the viral genome detectable. Using immunochemistry to identify astrocytes (GFAP+), oligodendrocytes (MBP+), and neurons (neurofilament SM312+), and laser capture microscopy to dissect out these specific cell populations, the authors were able to show that JCV DNA was found only in oligodendrocytes and astrocytes but not neurons. These studies clearly indicate that JCV DNA corresponding to all the major regions of the genome can be detected in brains from immunocompetent HIV-negative individuals. The fact that all patients had detectable JCV DNA in brain and that in one patient every region of the genome was also present, certainly suggests that ‘neuroinvasion’ may be characteristic of JCV virus infection even in normal hosts and not just a feature found only in immunocompromised individuals following reactivation from extraneural latency. The fact that at least one patient had detectable DNA encompassing all regions of the JCV genome also suggests that in some cases genetically complete latent virus capable of reactivation, and not just viral DNA fragments, might be present in brain. If these studies are confirmed and it can be shown by more sophisticated molecular biological techniques that the complete JCV genome is present in individual oligodendrocytes and presumably therefore capable of reactivating from these cells during host immunosuppression, the implications would be profound. In this scenario even sensitive methods of screening for JCV in blood or CSF would be unlikely to provide useful warning of the risk of developing PML.

How should these results alter our approach to studying the pathogenesis of PML? The possibility that strategies to screen blood or CSF for JCV may not be effective in reducing the risk of PML should focus emphasis on understanding the exact nature of the immunosupression that enhances risk of PML and of the normal host immune responses that prevent its development. In this regard it is important to stress the conundrum that regardless of the form of immunosupression, up to and including the presence of frank AIDS which remains by far the most significant risk factor for PML, only a small minority of presumably susceptible patients actually develop disease. The incidence of PML in AIDS remains <5% (5), and in patients treated with immunomodulatory drugs like natalizumab, rituximab, and mycophenolate is likely orders of magnitude lower (14). Despite having had half a century's worth of experience with PML it is discouraging that there remain no therapies or other interventions proven to alter the course or prognosis of the disease beyond reducing the degree of host immunosuppression. However, recent increases in understanding disease pathogenesis including identifying 5HT2a as a viral receptor on oligodendrocytes, and enhanced insights into JCV-specific immune responses, have at least suggested some intriguing new directions (15, 16).